I don't think you correctly interpreted my point, so I'll try to make it clearer. I don't think they will spend past rationality. I think they will be rational-- that is my point. As a result, I think they will spend as much as the expected increase in revenue. The expected increase in revenue, however, is NOT fixed, and the expected cost of talent is not either. The expected increase in revenue is based largely on the probability of spending putting a team in the playoff spot, something that has been shown many teams here and other places. Since teams have been tighter packed, that probability will go up, which will make the expected revenue of spending go up for teams on the margin, causing them to bid more on wins. You say that teams with 86+ wins will now have the option of spending more but you think they can choose to stand still. But that assumes is it equally easy to reach 86 wins. No one can realistically reach 86 wins without free agent talent, something I've shown before (look through archives for: Service Time Contracts and Wins, Part I and II), which means that if there are more bidders for that talent, the cost of that will go up. In other words, THE OPTION TO STAY PUT AT 86 WINS IS NOT AVAILABLE IF PLAYERS COST MORE. Another example, if there was no longer a luxury tax or draft pick compensation, the Yankees would spend more on the players that mid-market teams have been able to acquire in their pursuit of 86 wins-- making this cost more. The assumptions you're making assume constant prices of labor, a common assumption in perfectly competitive markets with many buyers and sellers of labor-- that has nothing to do with baseball. The odds of winning a playoff series is not so different than 50%, so they won't really treat that as different than before.

A fair point. Generally teams spend more than the direct playoff revenue. That extra revenue will be part of what the players get too, though, in the sense that they will be paid for the marginal revenue of wins. Whenever you add revenue associated with winning, the players will get that. So the net result here may be closer to neutral for owners, except for the bidding issue I mentioned earlier in the comments. So maybe it's not $50MM vs. $25MM but $50MM vs. $40MM. Something like that. Good call.

And regardless, it seems that you are undoubtedly convinced in general that you expect the players to be the beneficiaries. You're just not sold on my belief that the owners may actually keep less money. But you would agree that if these playoffs generated $25 million more in revenue that player salaries will go up at least around $25 million, right?

That's not what I'm saying. What I'm saying is that in situations where they are willing to spend $10 million, they used to only have to spend $8 million to win the auction, but now someone will bid them up $9 million. They will still make money but the increased competition makes it harder to get a player for much more than their marginal revenue. Not only that, when other teams spend less in general, there is less need to keep spending to ensure a playoff spot, but this may make it harder to continue to do so.

I played around with this one, but I think it's still better to ensure the longer series. The expected number of games in a 3-, 5-, and 7-game series are: 2.5, 4.125, and 5.8125. So you're expected number of playoff games as a wild card is 6.74 if you're in a wild card matchup and 8.48 games if you get a first-round bye directly. Definitely an interesting idea, though.

They don't necessarily spend up to their marginal revenue right now. Other teams bid them up as high as they're willing to go. But as more teams bid, they're more likely to spend closer to the marginal revenue and end up with the same chance of making the playoffs. The laws you're talking about implicitly assume that they would make the same amount of money if they sat on their hands. They wouldn't. Teams on the margin will spend more, and compete in bidding more. Also, the wild card/division winner issue in the AL East also comes into play for years with only two good teams in that division. Being the better of those two teams didn't used to matter but when your playoff odds as a wild card are cut in half relative to now, winning that division matters more. I don't know if you can assume owners will make more on regular seasons games proportionally just because they matter more. Maybe attendance goes up a little, but I doubt it would be commensurate with the playoff gains.

I'm not saying that going to 100% playoff participation will raise the marginal revenue. I'm saying that going from 27% to 33% will raise the marginal revenue, and specifically given how the league is currently set up. You're right that I'm saying that there will be more revenue in the collective pot, but that owners will end up with less of it. It follows logically from the economic concept that you're talking about. I'll give you a perfect example Suppose that the MLB says that any team that doesn't make the playoffs for five years in a row loses their revenue sharing. Let's say this makes fans excited, increases ticket sales. But as a result of the larger increase in marginal revenue per win to the Pirates, they now spend more. The revenue goes up, but the Pirates spend more money than the expected revenue increase. Contrived, yes, but it makes the same point. In this case, the odds of one game making the difference goes up from about 27% to about 43%. As a result, the marginal revenue of a win is higher to a lot of teams, and they bid up salaries. It follows logically from rationally acting owners who simply have higher estimates of expected marginal revenue from spending across the board. The key here is understanding the auction nature of this market.

The point I tried to make in last September's article was that this is about what PECOTA says, not what the authors say. I also really wouldn't talk about the Red Sox overcoming their sabermetric biases-- the other finding of that article (other than the fact that PECOTA overrated sabermetrically inclined teams) was that sabermetrically inclined teams did better per dollar than less inclined teams.

Every year this question comes up, and it's not a ridiculous one on first glance. However, you're missing the context of projecting stats of individuals, versus generic stats of leaders. We are NOT forecasting only two people to hit 100 RBI. If you asked for a projection of the number of people who will have 100 RBI, we can give you an estimate of 30 guys or whatever, but that doesn't mean we can give you which thirty guys will do it-- because what we can tell you is about 80 guys who are going to be at 90 +/- 20 with their true talent level. It's a matter of guessing what each player will do on average, rather than who will be the luckiest. If you give a pitcher with a 3.00 ERA a total of 33 starts, and randomize which games he pitches in, he'll win about 16 games on average. But sometimes he'll get 20, other times he'll get 12. We would be foolish to just start guessing which pitcher gets the best run support, just like we'd be foolish to start guessing which hitter has the most broken bat hits with men on third base. But of course somebody will get 20 wins most years. Somebody will get that lucky. Ryan Howard is the most likely to lead the league in RBI if you have to pick somebody, but he's only got maybe a 18% chance of doing so. Pujols has maybe a 16% chance, Braun has maybe a 14% chance, etc. We'd be foolish to sit here and say that since Howard has the highest odds of leading the league in RBI that he'd have the 130 RBI necessary to lead the league. There's a good chance he ends up with 90 RBI and we weight all the outcomes.

That's a good idea. I think the two categories you mention are important, but the unquantifiable intangibles include real skills too. I think things like coaching, synergy, etc., are all part of that. Once the new WARP statistic is available for all years, I'm going to be able to run that a little easier (and I plan to), and I can also check out things I do like the No Turnover Standings for last year too in which I follow the same sum-of-all-WARPs methodology you're talking about.

I kept switching between putting Francisco Liriano and Jered Weaver actually and ended up picking Weaver.

Yeah, I know, I just liked the alliteration. I guess I sort of meant the Maryland residents in the middle of the two. Bethesda is where I lived and, at least anecdotally, it seemed split down the middle.

The missing link here is that I've actually researched free agent performance in multi-year deals, and found that it declines pretty steeply after the first year: http://www.baseballprospectus.com/article.php?articleid=10883 And that's particularly true for older free agents. There are plenty of exceptions of course, but statistically the Nats are basically getting a bargain in the early years and a net loss in the late years. But the Nats would have been able to add someone who is in the "bargain years" of his multiyear deal when they will have the most use for a bargain. There isn't a lack of elite free agents in any offseason. The Nats just as easily could have signed the best OF or 1B in 2013 and not paid the premium of a current superstar that won't always be.

I don't underestimate the damage done to Washington baseball fans. I lived there, I saw it. It was nowhere near the psychological damage that Philadelphians saw when they had a decade of being called a small market team. I do not think that the Nationals will need to do anymore than the Phillies needed to do, which is win. And I'm not saying they will survive on Strasburg and Harper alone-- quite the contrary, I explained in the article that teams need free agent talent to supplement the product on the minor league system as it comes to fruition. But Werth is going to be a couple years older then, and it makes more sense to spending on the biggest free agent of 2013 (or whatever year looks like they're ready to compete) than to spend on Werth now and have less money to spend on players in their prime when Strasburg, Harper, and Zimmerman are all contributing.

The draft choices provided by Dunn belonged to the Nats regardless of what they did with Werth. I don't think that the Nats are going to draw much better this year with Werth than they did in the non-Strasburg games of 2010, do you? And I also don't think that an actual winner on the field in 2013 would be ignored due to fans being turned off in 2011, particularly if they added a top free agent in the 2012-13 offseason, do you?

Obviously just an anecdote, but I saw a lot of Orioles gear in Bethesda. I suspect if they were good, they'd have a much larger following, particularly if they sustained it.

Yup, I think that's the key here. It's not quite that simple because we need to factor in ticket prices and their effect, but the downside of fans distrusting your willingness to win just isn't THAT much lower than what the Nats and O's have been getting by spending on pretending like they're trying to win.

I'm sympathetic to your point, but I think that "showing the fans some wins" just isn't going well. These teams are selling 22K tickets per game, just 3K more than the Marlins and Pirates. I think that it's just not worth outbidding a team that is getting a combination of "showing the fans some wins" and actual playoff revenue.

I agree-- it's a lot of Maryland counties that are on the fence. I lived in Bethesda and it really seemed like an area that could go either way. I felt like I had a seat on the front lines, and neither army would get their act together!

Yeah, I wrote an article actually about when one-year deals work here (http://www.baseballprospectus.com/article.php?articleid=9869), and that actually does condone moves like signing and trading Tejada at times. At the same time, I still would like to see the money spent elsewhere in most situations. Especially with teams projected to lose 90 games or more, I think resources should be managed differently. It's a smaller loss, but why not use the $10 million that a Lee-type player costs towards international free agents and towards a free agent a couple years down the line?

I agree with the issue of having a city of transplants, as it is somewhat unique to baseball. Since baseball is a game that often is shared between fathers and sons, it is probably more likely to affect a transplant city where the younger residents share team loyalties with transplanted parents. Still, Strasburg-mania dulled those concerns, since I realized how many people wanted to be Nationals fans if they had a reason to be.

What the Nats got for Dunn and Willingham as irrelevant to this situation because they would have gotten those players regardless of whether they signed Werth. Similarly, the fact that they have Strasburg already shouldn't affect it either. By signing Werth, they lost the pick and that is a cost associated with it. However, you are correct that the Nationals have not been run as poorly as the Orioles have during 2005-2011, and that a lot of their under-.500 struggles are a result of the mismanagement of the Expos franchise. The only reason I highlighted their collective records was to discuss how unique the market was in lacking an above average team for an extended period of time. I agree that using free agency along with the draft is a smart strategy, but in a way that enhances the Nationals when they are going to be good. Free agents from other teams generally derive most of their value in the first year of their contracts. This is especially important for a team that is so positioned to improve over time as the Nationals are poised to do.

Right-- which I think is evidence of how much the area wants a winner! There are Caps jerseys all over the place, despite the fact that hockey on the whole is not that popular in America. People are rabid about the Redskins for the first two weeks of the season too. It's just an untapped market for a winner!

Was this what you had in mind? http://www.baseballprospectus.com/article.php?articleid=11890 The systematic bias I found was that teams who use sabermetrics more tended to be overrated.

This is a common question. If you asked the question of "how many pitchers will get more than 15 wins this season on average?", the answer is that probably 5-10 will. However, there is no individual pitcher with better than a 50/50 shot of topping 15 wins. The pitchers that win 20 games are basically always among the top pitchers in the games who were ALSO lucky with run support and health in a given season. For example, batting average generally has a standard deviation of about 20 points for a given season, which means that if you get 10 guys who are supposed to be around .310, one of them is probably going to hit .335 or so because of luck alone even without improving their talent level.

That could be something we'll work on soon too. I agree that it's important to be forthright about when we do not come out on top. Colin has worked very hard on improving PECOTA in the last several months, and he's definitely playing with tests of various versions of PECOTA at least internally. I do think multi-year looks at projection systems are an important step to take as well. Thanks.

I don't think we'll integrate the two in a formulaic way. Mike Fast and I email and tweet to exchanges ideas sometimes, and it has definitely helped my understanding of how the two can be integrated to learn about pitching. I suspect that the data won't be used in SIERA but both will be used to inform our understanding of pitchers.

Really cool idea. I'm going to have to look into this. Thanks!

My understanding is that is definitely the plan for the player cards this year. Doing it on old cards was apparently very tricky.

Yeah, I agree that FIP isn't park-adjusted and that causes it to be short-changed. In the article I explained that, which is why I tested FIP against unadjusted ERA for pitchers who did not switch teams, and checked SIERA, xFIP, etc. against park-adjusted ERA for the same pitchers who did not switched teams. (That's in table 4). I agree RA would be ideal instead of ERA, for non-fantasy purposes. At some point, we might derive a SIRA regression, or something like that, so as to sort out what that should look like. I guess that the original goal was just to move closer to estimating something familiar and to justify its use for fantasy purposes if people are interested. It's definitely something on my radar. Thanks.

In this article, I didn't actually compute any of the estimators using formulas. In the article, I explained that to avoid last year's mix-up, I just used what was available at BP and FG. I did use old SIERAs from 2003-04 before BP stats have access to the batted ball data. If there was a bbFIP posted somewhere that I could merge with the rest of the data, that would be easier. If I try to compute it from an original set of batted ball data, I can include bbFIP, but it's tricky without since my coding skills are still a work in progress.

Most of this stuff I looked at in my previous articles on this stuff actually. Great minds think alike and all.. Crowd seems to be almost no effect. If so, some teams would have consistent home-field advantages year-to-year more and more. This is not true. Also, teams that went from old stadiums to new stadiums would see jumps in HFA, but instead they see small declines if anything. Travel period seems to have a weird effect that is not very clear. There is no difference in between the first and second series in a row at home or on the road. There is a higher HFA in the 2nd game of a three games series, with lower HFA in the 1st and 3rd games of the series. This seems like it could be related to sleep deficits counteracting the effects of getting used to stadiums, but the jury is still out on it. On the other hand, travel length is very relevant. The shorter that teams travel, the smaller the home-field advantage. Umpiring I'll need to defer to others on. If you want more details, click on the link in the first sentence of this article, which is the recent article with links to all five articles in my 5-part series last year.

FYI When recreating MORP, I used the actual pick surrendered. The $8/5/14MM I talk about for Surrender 1st/Surrender 2nd/Resign Type A is just an approximation of the output. I literally went through each and every draft the last few years and wrote down which picks were surrendered. It was boring!

MORP is $5MM/WARP. So 2.47 WARP is $12.35MM. I'm not sure where you got the other number. (FYI the glossary needs to be updated from 2007 numbers which were based on 2007 salaries and on old WARP with a lower replacement level.) The 2.47 WARP would be right in line with a pitcher who throws about a 4.25 ERA over 180 innings, so that's where we're in agreement. If you recalculate with MORP done correctly, I think you'll realize where your mistake is. But Carl Pavano, a very comparable pitcher, is going to do a lot better than 2y/$17MM on the FA market right now. If you get in a bidding war with another team for Joe Blanton, you should go as far as you'd go for Carl Pavano. The first thing I said after the table was: "It is true that Blanton is probably the worst pitcher on this list, because it was designed to figure out who had outpitched him over the last two years". So of course he's the worst one in that list. Go through and find a list of pitchers who made $5MM after reaching six years of service time in the last few years. They all have major injuries or ERAs around 5. That's simply not Blanton at all.

Click on the "Statistics" tab on the top, click on "Pitcher Season - Rates" which takes you here: http://www.baseballprospectus.com/statistics/sortable/index.php?cid=112182 You can also click on "Pitcher Team - Rates" if you want pitcher's who switched teams mid-season to come up separately for two different teams (i.e. Roy Oswalt as an Astro and as a Phillie on separate lines)

That's a very specific cohort. It includes Joe Blanton and Johan Santana. Take away the service time minimum and you include Jason Hammel. Replace the IP of 350-400 limit with a 300-450 and you get Chris Carpenter.

I think what he is saying is that any variable that has any slight persistence at all will do that. That's true, I think. But we're talking about saying suppose that all pitchers gives up infinity fly balls in 2010 and infinity fly balls in 2011, all against the same competition, the correlation of his HR/FB in those two years would be 1.0. That's technically true. The sample size issue that Tango is talking about with HR/FB is cancelled out somewhat by the fact that HR/FB is lower than BABIP. HR/ofFB is around .130 on average. BABIP is around .300. About 28% of BIP are oFFB. So if you take a pitcher with N BIPs and .28*N ofFBs, then your standard deviation due to randomness is: For BABIP: sqrt{(.300)*(1-.300)/N} = .46/sqrt(N) For HR/ofFB: sqrt{(.130)*(1-.130)/(.28N)} = .64/sqrt(N) So there is a slightly larger standard deviation due to randomness, about 1.39 times as large for HR/oFFB. But the correlation for BABIP is .122 vs. .075 for HR/ofFB, which is 63% larger. So we're still looking at BABIP having higher variance in skill level (***assuming I did all that correctly***)

Well, the proof mostly comes at the beginning. The correlations of BABIP and HR/FB are positive and small. They are also significantly negatively correlated with K%, and BABIP is significantly correlated with GB%. So we do say evidence of that small amount of control in that respect. Pitchers who miss bats also tend to induce weaker contact.

Yeah, it's probably very unlikely that many pitchers can really control where hard hit contact is made. If you're going to have that much control, you should probably aim to get the guy to make weak contact or miss. But a few pitchers probably do have the skill and intelligence to get that job done, which is why we do see somewhat positive correlations for a lot of the numbers above, even the ones close to zero.

I don't want my discussion of FIP to be mischaracterized. When I say "FIP assumes no BABIP control," I do not mean "Tom Tango assumes no BABIP control." I'm ONLY saying that his model is designed to neutralize BABIP, and one of the advantages SIERA has is that it neutralizes BABIP luck with neutralizing the whole thing. One of the model assumptions of SIERA is that all we know about BABIP can be explained by K, BB, and GB%. That's not a 100% true assumption either. But it's a feature of the model.

I think that any detailed statistical analysis of one particular game needs to have a big asterix on it. That said, I think that pitchers (particularly ones like Moyer) are great at having plans for games. They are really good at mixing pitches and finding hitters weaknesses. As a result, they can often have good BABIP in a fleeting moment, but those strategies don't tend to be persistent. It's tough to keep getting the same hitters out if you are not striking them out or getting them to hit ground balls. Moyer will surprise a lot of hitters with a fastball down the middle on an 0-2 count, and then they'll hit a pop-up when they suddenly realize that it is actually a fastball. I don't think that Moyer will get the same guy to do that again, so I wouldn't say his BABIP is repeatable (as evidenced by his ever so slightly good but not amazing career BABIP), but he's definitely able to make a good pitch to get an out in an individual game. One thing I have also noticed with Moyer is that when his location is off, his BABIP is terrible, and no pitcher can always have their location on. It's a fleeting thing-- even Cliff Lee and Greg Maddux have games where the ball isn't going where they want it to be within the strike zone. So I would say it's possible to have a good game with mediocre K/BB, but probably it's rare, and the times it does happen, it's probably when the pitcher gets ahead in the count a lot.

Line drive rate for major league pitchers has almost no persistence at all. Something along the lines of <.01 year-to-year correlation. I'm sure that if you included a bunch of A-level pitchers in the major leagues, they'd allow a lot of line drives, but among MLB pitchers who can keep their jobs (i.e. maintain at least a K% of 10% and do at least something else well), they're line drive rate is not persistent at all.

I think so. They should be getting overhauled soon, and I've asked to make sure SIERA is on them. In my spreadsheets, I find it helpful to have SIERA year by year for pitchers so I'm sure readers would too!

Statistics tab-- Pitcher-Team-Rates or Pitcher-Season-Rates: Here's the link to pitcher-team-rates http://www.baseballprospectus.com/statistics/sortable/index.php?cid=224511 is Pitcher Team Rates

I didn't actually run this through peer review. A lot of the time our writers email each other about articles and talk them out, but short of a little discussion with Eric, I felt comfortable presenting these as facts. The conclusions are pretty evident, so I wasn't too worried about interpreting them incorrectly here. There's been some discussion of some of the peripheral issues on Tom Tango's blog but nothing really about the content at this point.

Oh, that's specifically because fly balls and pop-ups are outs more often than ground balls. About 23-24% of ground balls go for hits, while about 17-18% of fly balls and about 2% of pop-ups. FB pitchers have fewer hits on balls in play, but those hits do tend to go for extra bases, and they also tend to allow HRs too.

They each command the strike zone. Very few walks and plenty of strikeouts. You can be a slow baserunner as a hitter if you have power. You make up for the skill elsewhere. This is a similar thing.

Extreme fly balls pitchers are going to struggle to give up a lot of home runs. They may be solo home runs more often due to the infrequent hits on balls in play, but they will still be home runs. Pitchers who give up a lot of pop-ups are bound to do a little better than pitchers who give up only long flies. Regardless, the pitcher would need to have great K/BB numbers to offset the home run issue. There definitely could be a few. I think guys like Ted Lilly, Scott Baker, Jered Weaver all get away with high fly ball rates.

The p-statistics being low is good. It means that there is virtually no chance that these variables would be so far from zero if those variables had nothing to do with ERA. Specifically, it's very unlikely that the effect of GB% on BABIP is just linear. The curve that you see in the tables (and hopefully the graphs...they're working on it) is significant enough.

I'm actually discussing this in the next article which should be out tomorrow. There is actually somewhat of a skill with inducing pop-ups, probably related to movement.

Yes, there is a benefit to being a high GB% pitcher than xFIP doesn't calculate. xFIP assumes BABIP skill is equal for all pitchers so it misses pitcher differences in that skill. xFIP's strength is that it more precisely knows the direct effect on runs of HR, BB, and K than SIERA can tell its own. The tables aren't a takedown of xFIP. They're just a way of highlighting the effect of ground balls that I have found. The standard deviation of pitcher BABIP skill is about .007, meaning that it's probably about 0.15-0.20 runs per nine innings for the average pitcher. SIERA picks up on this skill pretty well-- and specifically does so because it has a GB^2 term in the equation. The point of the article is to explain why that term came up as it did, and how helpful it is to understanding pitching.

What I'm basically saying is that a 1% chance of such an extreme three-year period happening by random chance is still a chance, and that I'd bet this is just a fluke. I'm not saying I don't trust the numbers-- I calculated them myself. I'm saying that there's still a chance that it was a random fluctuation, and that I'm guessing that's what it will end up being. I'm sure that there is some hitter who is hitting over .500 in Tuesday day games, despite how improbable that is, and I don't think he's actually an over .500 hitter. But 1% of hitters will be 2.5 standard deviations above their true talent level in Tuesday day games, which would yield false positives. That's my best guess about what's happening here, but I'm very unsure. I don't think HFA and R/G are correlated at all, given that the numbers have barely moved for 60 years. Also, the R/G has fallen since 10 years ago but risen since 20 years ago, so we're not in a unique period in terms of R/G.

Z-score analysis on one-year samples requires a 56% Home-Field Advantage (up from 54%) to show up as statistically significant. There simply are not enough games in one season to confirm a switch to 55-55.5% Home-Field Advantage either way, so you will always discover no effect. It seems that you also may have used the standard deviation of the data, which is not the way to analyze this at all. We don't need sample standard deviation-- this is a binary variable. The standard deviation is sqrt(p*(1-p)/n) in any set of n games with a HFA of p. I think that using 3 years is going to be the best way to look at this, since we're asking about the last 3 years, and it's worth looking at 4 or 5 (like I did in the article) just to check if I'm cherry-picking data. You also would have found a significance in the 8-year period of 2003-2010 had you used the correct standard deviation formula rather than the sample standard deviation.

Hmm...I know that I tried to look at whether hitter's or pitcher's park would trend towards larger or smaller HFAs, but found virtually no difference though maybe a small tendency towards pitcher's parks. I'm not sure that I even thought about checking extreme vs. neutral parks...worth looking into also, thanks.

You know, I was looking for this effect a year or so ago, and it seemed like rosters were just built very poorly for their parks. But maybe that is starting to change more. That's probably just as likely as the quirky stadium effect, come to think of it. Worth looking at-- I just need to think about methodology because that's a tricky question to ask the data.

Maybe, but this seemed to be an improvement across the board, so I doubt it. It seems like a lot to be explained by cheating anyway-- even if every team stole signs and swapped out balls every single game they played, would it really change the outcome of 1.5% of games? Now suppose that only half the teams did it and only half the time-- still a lot of cheating-- would that 6% of games from losses to wins?

Yeah 1978-80 at 55.2% was as close as it got, but no three year period had 55.5%, so it's slightly unprecedented. And it is significantly different from the 1950-2010 average at the 99% confidence level. It's just that it still could be a coincidence, and I think that's still the most likely explanation. I think after 2011 and 2012, we'll have a better sense of this.

Yeah, I agree that it's not a very relevant correlation given the sample size. I only thought to look at it because of what I learned about the ratio of triples/doubles for home teams, and how stadium quirks would seem to exacerbate those type of difficulties on the road.

Okay, so firstly, I'm 100% sure that you are right that the better the performance of a player's teammates, the more likely a similar performance will lead to an MVP Award. I'm also 100% sure that the luckier your teammates were, the better there numbers were. However, I'm not asking that question. What we do know is that luck between teammates should not be correlated at all. If we define luck as we generally do-- randomness-- then we should not expect a correlation between teammates' luck. If Albert Pujols flips a coin that comes up heads and then gives the coin to Matt Holliday, his coin should still have a 50/50 shot of coming up heads. If Pujols' true talent level is a .400 OBP guy and Holliday is a .380 OBP guy, then Pujols getting on base more than 40% of the time should not affect Holliday's likelihood of getting on base more than 38% of the time (excluding things like IBBs).

I'm not sure I understand what I would change. Is your belief that the luck is correlated between players on a team? If so, I wouldn't call that luck. Maybe you mean that the things that tend to cause unsustainable jumps in performance are often common among players on a team? Maybe so, but it would be tough to quantify that I think, but I'm open to suggestions. Thanks.

I'm actually inclined to include Bonds, because it's not that surprising that one player every couple decades will exceed their competition by such a large margin. I think you want to include outliers here. Though for the curious, the WARP change would now be from 5.0, 7.5, 6.1 WARP, and .299, .327, .310 TAv. So it would still be about the same with a little more of a pre-MVP to MVP jump and a softer MVP to post-MVP decline. Bonds' 2005 injuries cause a 10.2-win drop in production that was atypical but his strong 2000 performance made the average improvement in MVP season look smaller. Still, I think we should include Bonds. Sometimes MVPs are won by Ruths and Bonds and sometimes by Vottos and Hamiltons.

D'oh. Fortunately, that mistake was made copying from Excel into Word, so the averages were done using Santana's correct statistics. His 2003-05 numbers were: 3.4, 8.6, 5.9 WARP; 3.07, 2.61, 2.87 ERA; 110.3, 228, 231.7 IP. Thanks.

Unless the market turned in the other direction. The average value of a win from a player with more than six years of service in 2009 and 2010 is already built into MORP, so the fact that outfielders were particularly cheap and other players were more expensive might not have been obvious. I'm not sure that there was conclusive information about which direction the market would go, and the fact that prices fell doesn't necessarily imply that the process was bad. Keep in mind also that Ibanez's performance might seem to have been a shock to us, but for people inside an individual baseball organization, they are privy to a lot more information. Given the familiarity of the Phillies with the Mariners' system, it wouldn't be surprising if they felt they had enough information to know he would beat his PECOTA projection. That he did beat it isn't necessarily conclusive about the process any more than knowing that prices fell is conclusive about the process in the opposite direction, but it's another thing to take into account. The net effect is that Ibanez produced more per dollar than most dollars spent on free agents over the last two years, even if you imagine the entire contract being spent on the previous two years.

Sorry-- the glossary definition is old. The descriptions came here: http://www.baseballprospectus.com/article.php?articleid=10629 and here: http://www.baseballprospectus.com/article.php?articleid=10642 The numbers for valuing draft picks were later put at $13MM for signing your own Type A, $8MM for surrendering your first round pick to sign someone else's, and $5MM for surrendering your second round pick to sign someone else's. Otherwise, the numbers in that article should be correct.

Ibanez had 4.8 WARP in 2009 and 3.6 WARP in 2010. That is worth about $41.5MM right there. He cost them $31.5MM for the three year deals plus a draft pick worth about $8MM. The deal is already about a wash if he is replacement level this year. Most multi-year deals are net gains early on and net losses later on.

Wow, I didn't even think of relating that article to this. That is definitely worth checking- thanks!

This is definitely vindication of that method, and even more so because the Hit List tries to adjust for strength of schedule. The Orioles do have to play in the AL East every year, so even if 3rd Order Record overestimates their future performance, it more correctly states their value.

SIERA does not account for league differences, so that's part of the reason why I'm inclined to say Lee is better than Lincecum. At the same time, Lincecum facing pitchers 2x per game probably wouldn't have an overwhelming impact. Some of the league difference in skill level comes from BABIP differences both on offense and defense, so I'm inclined to say that maybe a quarter of a run difference in skill level might be reflected in peripherals, but that could be off. Even still, saying Lee is better than Lincecum is not a statement I could make with 100% certainty, so perhaps I should dial that back a bit. In my opinion, I'd certainly pick Lee over Lincecum in terms of skill level but both of them are fantastic. Tonight should be a treat!

I discussed Blanton's 2010 season at length here: http://www.baseballprospectus.com/article.php?articleid=11954 Firstly, I would strongly advise you not to use line drive rate to measure pitcher performance. The correlation between a pitcher's line drive rate relative to his team is 0.003. Strikeout rate is the most persistent pitcher statistic, which is why it was not surprising to see Blanton maintain most of his strikeout rate this year. Also worth noting is that Blanton had a 3.48 ERA in the second half with a 20% K-rate, correcting some of the flaws he had in early 2010 with being hard hit. My argument for Blanton not being the kind of pitcher to pitch behind his SIERA is not simply the career history of doing so. It is also that he is a very normal pitcher. He doesn't have a knuckleball or anything that confuses these things. He doesn't have a bizarre pop-up inducing tendency. He doesn't have major splits within bases empty and men on. He doesn't throw very fast or very slow. He throws an average fastball at 89-90mph, mixing in primarily a changeup and a slider with an occassional curve or cutter. He's got years of matching peripherals and now has been able to inducing more whiffs than he did in Oakland. He's a pretty classic case of being hit a little bit harder and more balls falling in, but all in a non-repeatable way. And it hasn't been repeating since the All-Star Break, when I started chirping that he was going to improve. (Now cue jinxed Blanton having a small-sample size implosion in 4.5 hours...)

Bumgarner had a 3.88 SIERA this year, and Blanton had a 4.01 SIERA this year. Blanton walked 5% of hitters, Bumgarner walked 5% of hitters. Blanton struck out 18% of hitters, Bumgarner struck out 18% of hitter. Bumgarner had a 46% ground ball rate to Blanton's 43% ground ball rate. Their talent level is very similar right now. The point of SIERA is to isolate skill level from luck, and I have generally explained the good luck for pitchers with ERAs below their SIERAs and bad luck for pitchers above their SIERAs. In cases where I did not believe that was appropriate, I highlighted pitchers like Matt Cain who seems to have a unique skill at beating his SIERA. I'm not convinced Bumgarner has that. He will probably improve going forward, but that doesn't make him a decisively better pitcher in today's game. Both are very good 4th starters, so "one of the best fourth starters" would be equating his skill level for Bumgarner's, generally acknowledged as a good fourth starter.

Check Blanton's SIERA the past two years (3.92, 4.05) and let me know if you think that something about Blanton's pitching that makes him unable to match his SIERA. I don't see any reason to think he's something other than a 4.00-4.10 ERA pitcher when it comes to true talent level. Keep in mind his ERA last year was right on with his SIERA, as it has been in years past. Bumgarner is a great fourth starter as well. Both are leagues better than the majority of fourth starters in the league. How many other teams have four starters with talent levels of 4.10 or better?

The indented paragraph is the re-printed summary of what I wrote for each pitcher's first start. The analysis of the season/career of the pitchers still holds, and it seems silly to rewrite the exact the same thing, though maybe I'll eliminate the references to LDS opponents going forward if it's confusing.

I meant that he gets two games allotted to him. If the series ends first, it won't happen of course. The Rangers still need at least 2 wins from non-Lee pitchers.

In the sense that it affects sequencing, sure. Tom Glavine notoriously bested estimators like SIERA in his career because of how he sequenced-- specifically, attacking hitters with bases empty and pitching around them more with men on. The result was more of his opponents' OBP came when men were on (and SLG was relatively more important), and more of his opponents' SLG came when bases were empty (and OBP was relatively more important). His UBB rate went up from 6.2 to 8.9 percent with men on, while his HR rate went down from 2.3 to 1.5 percent. As a result, you get things like: 2005: ERA 3.53, SIERA 4.72 2006: ERA 3.82, SIERA 4.37 2007: ERA 4.45, SIERA 5.20 That's not quite what most people mean by mental makeup, but it's mental in the sense that he was an excellent strategist with good control. In terms of "bearing down" which is what most people mean by mental makeup, I think that it could work the other way-- certain pitchers with anxiety issues may perform worse with men on and have ERAs worse than their SIERAs, though I suspect people with that tendency might not have buckled down when they needed to pitch in front of scouts the first time anyway. I also think that people who don't focus with bases empty probably would have trouble making the big leagues too-- so they might have a tendency to have ERAs lower than their SIERAs but both would be too high in the first place. I think people imagine pitchers and hitters out there with intense concentration at crucial moments in the game, but I tend to think that's just when fans concentrate really hard and notice the concentration a pitcher or hitter that was concentrating hard the whole time.

SIERA does not factor in run support at all. It focuses on what a pitcher's ERA should be if he had neutral luck and played in a neutral park with average defense. To your question about best rotations in the playoffs, the Phillies rotation is better than the Giants in terms of SIERA, with three of the top six SIERAs in the playoffs (among starters). Even if you adjust for Cain constantly being able to beat his SIERA, it would still be in the Phillies favor as Sanchez is a good deal behind Oswalt and Hamels. Keep in mind Blanton as a 4th starter is much better than Kendrick-- his SIERA this year was actually just 4.01. Still, I think it's probably safe to say that the Giants have the second best rotation in the playoffs, though the Braves and Rangers both have pretty solid ones too.

I'm guessing that the park adjustments would very a lot from pitcher to pitcher, though, because different pitchers throw in different regions with different angles more than others. A guy with a sweeping slider might be particularly tough to evaluate. On top of that, my home-field advantage studies found that the strike zone was where large home/road differentials existed, with BBs and Ks being big areas of home-field advantage. I would think adjusting pitcher to pitcher could be tricky. All in all, I think doing it without a scientific calculation like pitchF/X is going to be problematic. I think it's a great approach to baseball and the results above show it does have some value even with all its problems, but the granularity of noisy information makes it tough to really approach scientifically and learn something about.

Hmm...I will look into this tonight when I have access to my data again. The first-pitch strikes was about the year before though-- the high walk year-- so the pitchers who had thrown more first pitch strikes but had higher walk rates anyway improved their walk rates whether or not they had more first pitch strikes the second (lower BB) year.

You completely misquoted him. DL asked Burns that question while following up with a comment on whether it was polarizing. It's clear from Burns' response that he meant that it mirrors society a lot (which is his whole point anyway), but that the polarization of society is not quite mirrored in baseball. Separately-- The issue of subsidizing stadiums is whether it generates enough money to give the government more money to spend, so you're not focusing on the right issue there. Also, the value of seats is not a moral issue, and shouldn't thrown in either. A price ceiling on tickets would just create a black market where scalpers would be the beneficiaries of baseball teams' and players' hard work.

If you read J.C. Bradbury's article on aging, he found that strikeouts peak so early that having a normal aging curve wouldn't have been appropriate. The correlation between SO/PA and the other variables is very high, which is why I needed a regression in the first place, to see which effect was driving it. They aren't colinear or anything, just highly correlated, so it's appropriate to run the regression with both K% and the correlated BIS variables like swinging strike rate.

Weird, I somehow mistyped the tables! The one labeled Swing% is O-Swing%, the one labeled O-Swing% is Z-Swing%, and the one labeled Z-Swing% is Swing%. O-Swing% was weakly statistically significant.

I think the last correlation you found is a result of selection bias. There are pitchers who are good at getting ground balls, pitchers who are good at striking guys out, pitchers who are good at both, and pitchers who are good at neither. Halladay is good at both. Hamels is good at Ks. Hudson is good at GBs. The guy who is good at neither is in AAA. Hence you get negative correlation. I'm not sure that high strikes are the best explanation, because when I ran my regression in the spring in the "Why SIERA Doesn't Throw the BABIP Out with the Bathwater" article, I found that even controlling for ground ball rate, high-K pitchers had lower BABIPs. Controlling for Ks, high ground ball pitchers had higher BABIPs. I think both are distinct effects. One is about the types of batted balls most likely to go for hits (GBs vs. FBs) and the other is about batter's decisions with how hard they swing, etc.

I definitely have found the BABIP vs. SO% inverse correlation and written about it some before. I would suspect that is stronger than the swinging strike vs. BABIP correlation, but it's worth checking at some point when I can merge all that data in one set. Good idea, thanks.

It's true that we don't have the balls in play data, but it's pretty clear Ted Williams was lucky that year because he hit .344 the year before and .356 the year after. So either he (a) became ridiculously more talented than he already was for one year and lost that talent immediately afterwards (b) faced ridiculously poor pitching/defense that one year in such a way that his Three True Outcomes numbers didn't move (c) figured something out very amazing about pitchers that one year that completely did not affect his Three True Outcomes numbers and which pitchers immediately fixed the next year or (d) he was a little lucky that year. So with a little bit of a cop-out, I'm gonna say he was lucky, but yeah, no batted ball data means no E-BABIP.

Well, the .367 BABIP is at least partly luck. His minor league BABIPs weren't that high, but were all pretty consistently above average. His K and BB numbers certainly don't indicate he controls the strike zone all that well, and he only has a few HR, so this isn't a guy who really drives the ball. What he does have is a slight downward plane to his swing which is giving him a 50%-ish ground ball rate and few infield flies. That's why he's above average at BABIP in terms of skill level, especially when mixed with his speed. I'd say he's probably a .315-BABIP kind of guy, but I haven't pushed him through the model yet to see what it would say.

With Young, I actually had him projected at .316 BABIP for 2009, so .313 seems about spot on. I'd need to look at his numbers more carefully to say what has changed, though, and it would take some time to calculate his E-BABIP for 2011. I had to plug the other guys in individually since I didn't have a whole spreadsheet put together. I'd guess around the same area though. His line drive BABIP was certainly high in previous years, which usually suggests regression though. I'm guessing that's a lot of what happened.

With random variance formula-- it is highest when p is closest to .5, which means ground balls (.456) will have a HIGHER random variance than line drives (.189). It really depends on the relative noise to skill. For instance, home run rate is further from .5 than BABIP is for hitters, but it has a much higher year-to-year correlation because the size of the home run skill across major leaguers varies so much more than the variance of the BABIP skill.

I'm not really sure it has to do with sample size, because it's still the same sample of balls in play. In fact, the standard deviation is lower for a low percentage outcome. If ground balls are 45%, than their standard deviation over 500 balls in play would be 2.22%, but with line drives at 19%, their standard deviation is 1.75%. It's about the relative skill level standard deviation being much smaller. There are guys who hit 55% ground balls regularly, but there isn't anyone who hits 29% line drives regularly. I wouldn't even think it's measurement error because that should actually make unadjusted line drive numbers more persistent because the bias is more persistent. When it comes to line drives for pitchers, they look a small bit persistent only before you look at line drive rate relative to team and than persistence falls to 0. The only reason line drives appear to happen more often is that scorer bias is persistent. The reason I think skill level with line drives is smaller has to do with swings. If you're good at putting your bat on the ball, you'll have more batted balls, but I'm not sure you center it that much more regularly. However, if you swing with an uppercut or a downward plane, that is going to regularly come off the bat a different trajectory.

This is somewhat new for him this year. His career rate of ground balls going to the outfield is just 16.2%, near the league average of 15.3%. In 2010, it jumped to 20.0%. Is he hitting ground balls a lot differently this year perhaps?

It is random from the pitchers point of view, not the hitter's. Hitters with more power regularly hit the balls hit to the outfield over the wall, but pitchers don't. After adjusting for park, the correlation between a pitcher's percentage of outfield fly balls hit for home runs one year with the the same percentage the next year is only .08. It's almost entirely random.

Oops. Yes, you're right. Thanks.

This is a very intriguing hypothesis, and I don't think the CERA studies have much to say on whether catchers are better matches for certain pitchers, just that catchers don't tend to be repeatedly much better or worse than each other. That doesn't really comment on catchers sucking at calling pitches for a certain pitcher. There is abundant evidence that Shields is pitching worse when Jaso is catching. K% down 2%, BB% up 1.5%, XBH% up about 5.5% while singles are only up like 1% or so, definitely suggesting that he is being hit harder and throwing worse with Jaso behind the plate. My question is why? Is Jaso calling games very differently? Is it possible Jaso is just getting more playing time against better hitting teams? If it's game-calling, what is Jaso doing wrong? That's the question I'd like to see answered. I don't have the PitchF/X chops to answer a question like that, but it's a very good question. If he's calling pitches predictably, that could be an obvious effect. Keep in mind that would be another example of high BABIPs not being repeatable but still not being attributable to simply bad luck. There would be an action that was bad-- poor playcalling-- that would be reversed by a team that was paying attention to what they were doing, and therefore there would low/no correlation in BABIP the next year.

Could be, but statistically that gap is just too wide to be very plausible-- two full runs! Also, keep in mind that Hudson still has been lucky and has a WAY more extreme rate of ground balls, making an extreme ground ball BABIP more credible. What type of sink is he putting on the ball to get 2/3 of contact on the ground? Is it making more choppers and fewer one hoppers through the hole? Regardless, Buchholz may be maximizing the value of the Red Sox defense but it won't cut his ERA in half versus what his peripherals suggest, in my opinion.

More or less, I think I agree with what you are saying but provided that we define DIPS somewhat loosely. I think that pitchers have SOME control over BABIP, but I tend to think that you can deduce their BABIP tendencies better by looking at their strikeout and fly ball tendencies. But otherwise, yes, that would be how I think about it. I do not think current defensive metrics are good enough to pinpoint the defensive adjustment, but I think that if you approximate the runs above or below average and spread that by innings pitched or something, you can get a good ballpark number. But I do think that the extremes with defense are going to be +/- 0.50 runs or so, and the extreme with parks are going to be about the same, so even a perfect defense in a huge stadium would have less than a run difference between ERA expected and SIERA for a pitcher. For the vast majority, of pitchers, the effects will be far smaller than a run. The Red Sox have pretty good defense but a hitter's park, same with the Phillies, so those effects might cancel out. Or at least they won't explain a large portion of the SIERA-ERA difference. For Santana, the park effect is going to be relatively small and I guess the defense is pretty average, so you can probably figure that a 4.18 SIERA in CitiField playing for the Mets defense would maybe be at 4.00ish? Maybe 3.90ish? Certainly not 2.98. Cahill I guess gets a double bonus for park and defense, but that would move him from 4.28 SIERA to maybe 3.5-3.6 expected ERA? Certainly not 2.72 or anything close to it-- there's just too many ground balls being hit at infielders even for the A's. I do think it would be ideal to come up with a way to translate SIERAs into "expected ERAs" or something like that, but defense is not so exact a science that we can trust it perfectly when we get numbers. I think when Colin Wyers published his runs above average collectively for teams, we can take that and the percentage park factors that Clay does, and then make a noisy adjustment. But that would really just be a guessing game at this stage. Good question, thanks. I'm really liking everyone's questions on this article.

Sequencing could definitely be an issue, and I agree that's partly a catcher issue. The league average pop-up rate is about 7.5% of all balls in play. The league average BAs with bases empty and runners on base are .253 and .266, respectively, with OPS of .715 and .755. 37.8% of two-strike counts have ended in Ks so far in 2010.

Definitely not a stupid question! I think you just gave me about four article ideas in three sentences! The answer is probably a great question for a scout, but the logical starting point is-- why aren't his swinging strikes happening on two-strike counts as often as other players? Is it a matter of learning to pitch? Is it a matter of having a few good pitches without an obvious putaway pitch? Or the opposite? I don't know. Worth hearing what scouts and regular viewers of Red Sox games have to say. Thanks for the fantastic question.

I'd guess it's probably related to just missing on talent level of players, which is certainly a flaw of sabermetrics. As far as OBP and K's, I think that sabermetrics has very carefully approximated how well those things contribute to wins, but might be missing on the OBP and K tendencies of the players themselves.

I think with most of these things, the only reason that there would be an issue is if teams that were less sabermetrically inclined were more likely to have the factors you have mentioned. From your wording, I think you realized this, but it should be highlighted. While PECOTA may be conservative on prospects, teams that were sabermetrically inclined would have to be less likely to develop have ceiling prospects in the first place. Otherwise, it would not affect the correlation. Similarly, good medical staff would need to not only be negatively correlated with good sabermetric staff, but the difference would have to not be reflected in historical injury trends, which PECOTA does adjust for. Synergy would also have to be better understood and better utilized by less sabermetrically inclined teams, which I'm not so sure is true. It's possible though. Defense is included in PECOTA projections in that pitchers on teams that are more likely to have good defense will have lower ERAs (and RAs). The issue would have to be that teams that are sabermetrically inclined would have to be less likely to pick up defensive players that only look good by the numbers-- entirely possible, and a solid plausible example of where the effect above might be seen. As far as the chance of this being random error, it's possible. The correlation is .27, which isn't massive. But considering how many other random factors there are, I would never have expected to get a correlation even that high.

This would only affect the correlation between being sabermetric and being good, not the overprojection issue, so it's only a limited effect. I'm still not quite sure that there are obvious examples of teams that were overrated as far as sabermetric usage. Boston really does seem like a good candidate for most sabermetric, you know? The teams that were ranked the highest are the ones that sabermetricians are known to be on staff at in many cases too. It's a fair point and it certainly suggests that these conclusions should be seen as qualitative for the most part, but it seems unlikely that it would change the results much. If you have some teams in mind that were possibly badly rated, though, please let me know.

Wow, this is a great post. I need to think about these questions. I don't have immediate answers, but this should definitely be filed as things to look at more deeply in the future. The managers question is awesome. The bullpen question is somewhat possible too, as the correlation between overprojection and pythagorean record is .22, lower than the overprojection vs. real record correlation of .27, though not by too much. The rest of those are good ones to think about. Thanks.

Checked out the correlation from PECOTA and 1st order Pythagorean Record. Still .22, but definitely less than the .27. Good food for thought. I'm not sure people were biased by PECOTA's preferences. I think that it's a tough leap to make that PECOTA liking a team means they are sabermetric. I don't know people would assume that a high PECOTA projected win total must have come from a secret saber genius at the helm. I could be wrong though. I didn't even fill out a survey because I was afraid I would bias the results.

There's some holes in these arguments that I think you're missing. Hear me out... 1) Potential bias ONLY affects the rating of how much saber teams help you, and does NOT affect the primary conclusion of the article at all-- in fact, it would have the opposite bias. It seems like my colleagues were very careful to rate teams highly in cases where they were not the cream of the crop. Although there could be some effect, I'd like to hear some team names first to even address that weaker side-conclusion that the saber teams were better. 2) Huntington was hired by the Pirates before the 2007-08 offseason, meaning 3 of the 5 seasons in question were with a sabermetrically leaning GM. My colleagues ranked them 10th. The Mariners were 5th even though there were only 2 years of Zduriencik at the helm, but they have been so extremely highlighted in the media for being saber leaning that it's not a totally ridiculous ranking. Overall, the Mariners did not significantly differ from their PECOTA projections at all (1.7 games overall) which is pretty much coming entirely from 2010, when PECOTA did, in fact, overrate them. 3) Opponent difficulty is adjusted for in the PECOTA standings and obviously real life standings are affected by it as well. Despite your focus on the AL West, there is actually a POSITIVE correlation with being sabermetric based on your sabermetric leanings. Look at the saber-tastic AL East and contrast it with the saber-lacking NL East. The top 6 teams in sabermetric ranking were all in the AL. 4) I'd be interested in this too, and that's a good idea for future research if I want to really get my hands dirty. A starting point could be some of my articles on "The Cost of OPP" and free agent PECOTA bias. 5) This is a pretty minimal effect, I'm guessing. The Indians are the premier example of this, overprojected by 10 games a year, so about 50 games overall. Cliff Lee and CC Sabathia were only worth 6.5 games to the Phillies and Brewers combined. The Yankees add a lot of talent and are actually listed as more sabermetric. Also, there is a slightly positive correlation between being competitive and being saber-leaning, so there is going to be a counter-effect whereby saber teams add talent midseason anyway. I do appreciate the criticism here, but there are major holes in a lot of these arguments. If you think the authors are wrong about the saber usage of a team, tell me which team it is. A couple teams won't change the conclusion but it is certainly worth hearing.

No, because each season is only one observation among an infinite number of paths a team could take in a given year. There is no absolute truth in one season or even five seasons of data. There is still a margin of error.

Thanks for looking into the regressions. The issue is that there are too few observations (just 30 teams!) to really interpret a multiple regression in any meaningful way. The correlation was .27 even ignoring the payroll or payroll rank issue, but that's not significant in a regression because of the number of variables. Including raw payroll is a problem because of the little outlier in the Bronx that has such a high deviation from average in both payroll and wins. More than 30 observations would remove issues like that. I did run a couple regressions to get a feel for the data, but didn't report them because of the sample size issue. Nonetheless, thanks for looking into the data.

They are all valuable in different ways, and there are also three versions of WARWARP. Two main distinctions are important: 1) Measuring outcomes vs. Measuring skills 2) Measuring cumulative performance vs. measure rate of performance FanGraphs WAR uses FIP, which is like SIERA in that it measures a skill, but unlike SIERA in that it measures how well you did cumulatively instead of as a rate of performance like SIERA (which approximates earned runs per nine innings). BP WARP and Baseball-Reference WAR measure outcomes as a cumulative performance, while ERA+ measures outcomes as a rate of performance. WARP and WAR measure outcomes relative to a replacement player, while ERA+ measures outcomes relatively to average. Hope that helps!

The typical boxscore in a newspaper typically doesn't write down batted ball types, but even FanGraphs boxscores just lump it all in with fly balls. That's okay because we have fly balls and pop ups always added together in the equation so it's okay to call all pop ups fly balls and totally have the equation work. Any game summary on Gameday will include pop ups too.

Okay, Eric's awesome league splits by situation spreadsheet he sent me actually can help see how true this is: Bases empty: LHB BABIP: .298 RHB BABIP: .293 Runners on 2nd & 3rd, with 1st base open: LHB BABIP: .291 RHB BABIP: .302 Runner on 2nd, with 1st & 3rd both open: LHB BABIP: .293 RHB BABIP: .292 Runner on 3rd, with 1st & 2nd both open: LHB BABIP: .305 RHB BABIP: .302 Runner on 1st, with 2nd & 3rd both open: LHB BABIP: .319 RHB BABIP: .306 Runners on 1st & 2nd, with 3rd base open: LHB BABIP: .295 RHB BABIP: .291 Runners on 1st & 3rd, with 2nd base open: LHB BABIP: .316 RHB BABIP: .305 Bases loaded: LHB BABIP: .303 RHB BABIP: .293 So it's definitely somewhat true from looking at these, I think. Worth noting above is that the BABIP with RISP vs. bases empty is only +.0115 for LHB and only +.0041 for RHB, compared to the men on vs. bases empty .0198 for LHB and .0059 for RHB, so maybe my 27 shiftees up there are doing better with men on 1st especially, somewhat better with other runners on, but not as much of a jump. That makes sense. I don't have their numbers by situation though.

BTW studes linked to the Greenhouse work below I'm seeing now.

I'm not at home so I don't have my copy of The Book to look through with me, but did they control for hitter skill? Could that explain the difference? Because lefties are better hitters overall than righties (I think I read maybe on THT a few years ago that had to do with more righties being 2B, 3B, SS, and C, rather than 1B and OF), maybe they face more same-handed pitching with runners on when managers bring in relievers specifically to get them out. Maybe the league as a whole faces a smaller advantage. Eric sent me all the BABIP numbers above and they matched up with the league numbers on B-R, so it should be accurate. And we do see a 1.5-point jump in BABIP for lefties more than righties with men on. I'm not sure what that means for wOBA, but it's certainly some fraction of that 20. I'm curious what you find. Actually, maybe righties hit better with men on 2nd and 3rd but not first because the 3B and SS can't play as far back? Maybe that makes the overall "runners on" numbers closer but the "first base" numbers much more advantageous to righties. That would make sense in explaining why the shift causes a wider gap because even with runners on 2nd and/or 3rd and first base open, it would still severely limit the shift.

Yeah, I've been told that a couple times today, though I really only was focused on the 1993-2010 era. I guess it's a matter of relative use. The righties above in the control group were similar to the lefties in the test group, except for alignment.

Yeah, that's why I looked at all lefties and all righties to check the BABIP difference. There was one but it was only 1.5 points difference.

LOL. I'd like to thank to academy...

Sure, teams could do it anyway, but once you have your first baseman out of position, you need to think about moving the second baseman over the cover some ground, and the overall result is just less optimal-- it's not like the difference in BABIP is 100 points. It's 20 points-- just enough to see the suboptimality of defensive alignments with runners on when there are goals other than minimizing the chance of getting a hit.

I think punching ground balls the other way does counteract the effectiveness, but I've read that it's very hard to put breaking balls on the ground the other way so maybe that limits the ability of hitters to do this. Also, forcing hitters to hit the ball the other way might lower their SLG even as it could raise their AVG. I think the overall effectiveness of the shift is somewhat unclear, though I do think this article would suggest it at least works a little. Jeremy Greenhouse has done some interesting work at BaseballAnalysts.com on this issue, might be worth checking out (though I didn't save the link...you can just click on his name on the sidebar and you can see all his articles).

Darwinism explains a wide variety of topics in economics-- both biology and economics inform each other well. I had a friend who did both and was able to utilize both fields to understand them both better. However, the survival aspect of Darwinism generally links to the entry/exit of firms condition in economics, which is not something that is relevant in Major League Baseball. Baseball teams do not go out of business, are not forced to dissolve when better competition emerges like restaurants are. However, the general point that teams from bigger markets stand to gain more from investing in winning is true, regardless of whether you frame it as they stand to lose more by not winning. In any case, the general point of the article holds. As long as the return on investment is vastly different in large and small markets, the large market teams are likely to outbid the small market teams unless the cost of investment is lower for the small market teams.

I agree with your point that the decision to spend money on a team is one of investment, not of charity, and based on building a customer relationship. However, I was assuming that the value of building customer relationships in New York City is higher both because of the sheer number of customers and because of their wealth, so the price of talent-- which builds that fan relationship-- is driven up by the big-city and/or playoff-bubble franchises with the most return on that investment. So I'm guessing that not participating in the auctions for players is probably rational for a small-market non-playoff-bubble team. I think you and I agree, basically, in principle though and maybe my language of "for the sake of making fans happy" was a bit of an exaggeration of the general fan sentiment of calling their owners "cheap."

This is all true, but it comes down to asking what the goal is. Do you want the Pirates to neglect being less terrible for the next few years so that they can be championship contenders after that, or do you want the Pirates to put a team on the field that is going to give the Cardinals and Reds a run for their money in pennant-deciding games? I'm pretty sure baseball might care more about the latter more than the former. But if not, sure, go ahead and subsidize development cost too. That's why I said "a number of ways" in the article-- I was hedging on that one!

Excuse me? I also fail to understand your point, Peter. I'm not sure what payrolls having a "complete lock on team income" means. I think it's also pretty clear that you seem to have come to the same conclusion as everyone else-- that building a winner is more profitable in large markets-- and then you seem to think I don't think that, despite the fact that the crux of my argument was precisely that...Oh, and I do own The Wealth of Nations. It was required text in the intro courses to my Ph.D. in Economics. I'm not sure I read anything in there about anything you said though. Why don't you try to see what you disagree with in my article, and we can discuss?

Yeah, something exactly like that. MLB contributes towards payroll at the low end of the payroll distribution and subsidizes it, while taxing at the high end. The "number of ways to implement" was just to say that anything that involves lowering the marginal cost below the actual cost of the salaries for low salaries is good. The reason it's okay to being effectively paying everybody $15 million is that a lot of these teams are receiving that money now anyway but not raising salary. You get the money from taxing the high end of the curve and from things like MLBAM. There could also be a phase-in payroll subsidy that gradually becomes a payroll tax, too.

Well, a 1-win player at the beginning of the season increases your playoff odds by about 6% so it would only be $1.8MM in expected value (6% of $30MM). However, there are two other factors. Firstly, there is also a dollar value of winning games independent of the playoff affect, and that probably makes up half (maybe?) of the value. Not only that, but the $30MM as Gennaro describes in the interview is for an average market-size team (he cites Cincinnati). Keep in mind that the team that offers that free agent the most is probably going to have a higher than average value for him-- often free agent dollars are spent most by teams from New York, Los Angeles, Philadelphia, Chicago-- all bigger markets than Cincinnati and probably with a greater than $30MM value if I understand him correctly. I'm just playing with the numbers, but for an average team, the $30MM sounds reasonable.

I do not have inside knowledge of packages but I explained my reasoning in several articles, including this one, why the value is likely much higher. There were rumors that their asking price was through the roof, perhaps explaining why they did not complete a trade. I moved to DC a year ago, so I didn't bus down, and I wasn't killing any kegs; I had a pretty good hot dog, though. I was treated civilly by fans that I treated civilly as well. The vast majority of fans in any city treat you well enough in their stadium if you behave yourself, and that's certainly true for DC and Philly.

Fair enough, but how exactly would I go about proving that? I'd need to try and capture a market that was different every summer and every winter based on a bunch of different factors. I could see why it might be hard to trade position players whose value comes from their place on the defensive spectrum, but probably not 1B/DH types.

I don't like the return for Oswalt + Berkman, but I just think the demand wasn't there right now because they were expensive, Berkman was overperforming, and Oswalt had a no trade clause. I think the economy and interest rates made it a bad seller's market, but also made holding expensive players really tough on sellers. I did think Happ wasn't the best match for them.

You aren't linking the PECOTA projections. You're linking the staff predictions. Mine, incidentally, had Cincinnati in 2nd, so that's not really a fair analogy. As I mentioned above, the Padres clearly tried something innovative that worked. And they still aren't a guarantee to make the playoffs, and they were able to spend some money midseason to correct things when they realized they were better than they thought. It's not a matter of giving up. You still play the games. You just allocate resources differently-- as in, you invest in 2012 when your odds of competing are three times as high, even if they are not zero in 2010. Also, Dunn is a free agent after 2010 so his contract for 2011 would have to be a new deal, and wouldn't be a bargain. So, considering they'd be paying full price for him, wouldn't it make more sense to do so when they already know they're competitive (or to trade for a first baseman midseason if they are suddenly in the thick of the race in July 2011)? Allocating resources is a big deal. I'm not saying the Brewers and Nats won't be good this year. I was wrong about the Padres this year. I'm saying that teams that have a lower percent chance of competing shouldn't spend money on winning now when that money could be used towards later. Similarly, they shouldn't hoard talent now at the expense of trading for talent that can help them later. Washington is a pretty big and wealthy city, and the Nats are going to be competitive if they are run wisely. A sustained period of success could come to them if they build the groundwork beneath them.

I'll say this. I'm a Phillies fan myself, and I watched through the late 90s and felt the same way you describe until 2008, when I was at (both parts of) Game 5, and realized every loss endured while the team was bad was worth it if it made that championship happen. I'd stay open to the possibility of changing your mind. That said, I never had the experience my cousin did: http://sportsillustrated.cnn.com/vault/article/magazine/MAG1161294/index.htm (btw, I'm the "number-crunching cousin" in that article).

I was trying to reply to a few of his posts at once, and he mentioned a couple times that he didn't like my snarky comment about the Nats fans at Saturday's game. Sorry that wasn't clearer.

I understand and very much appreciate you clarifying. I shouldn't make unnecessary snarky remarks like the Dunn comment. If you were sitting in my section that day, you would probably agree those fans knew far less about baseball than you certainly do.

BTW if you go to the first link in the article about buyers and sellers, I make a pretty detailed case on how to determine where the buy/sell line should be, and how large the range of prices which would work is. The teams with 1% or less change in championship odds are clearly sellers and the teams with 3% or more change are clearly buyers, and the 1-3% range in the middle is pretty small and those teams clearly are incentivized strongly enough to make major moves by either price.

The Padres clearly were doing something smart here and were the exception rather than the rule. They still aren't guaranteed to contend, and were able to make midseason improvements anyway rather than dipping into the free agent market. The Padres also played in a division where the Dodgers had not spent any money to maintain their performance, the Rockies were good but not necessarily considered a 90-win team at the time (though I had them very high), and the Giants and Diamondbacks weren't clear contenders. With the Phillies and Braves already looking pretty intimidating for 2011, it's not wise for the Nats to beat them now. The Phillies have a bunch of 30-33 year olds on deals that expire soon, and the Braves don't spent much money, leaving the Nats a real window if they aim for winning in a couple years, especially with having such elite players already in the fold and under team control. I just don't think the relative value is high enough now even if there is a 5-10% chance of making the playoffs in 2011.

Actually this is pretty close to what I was going to say for part of my reply, though maybe including the snark about Nats' fans was unnecessary. I generally try not to do that. Anyway, it's not really supply and demand in an intro micro sense. It's an auction market. The price is going to be determined by the next-highest bidder. There were a few 1B/DH types on the market, and more teams that could have used the boost, so the price was going to be determined by how much the next-highest bid would have been-- which is about value for two pennant chases, not one. And that team would likely have at least as good a chance as the Nats in 2011, so they would have more value for Dunn as well. There is not going to be a mathematically rigorous way to talk about these market effects. It's certainly more true with pitchers, though I do think the Johan Santana return is probably a great example of what happens when you trade with one season left. I've seen a few examples over the years where the return was starkly different but I'd need to think about this before I could even devise a method of coming up with a rigorous way to analyze it. After all, establishing prospect rankings would be very tough to do in a non-contraversial way.

Agreed. This is the reason that I didn't totally write off the wins Saunders generates-- it's an evaluation of the market value of the package with an agnostic view about which parts of the package are most important.

The problem with that logic is related to my articles on MORP where I determined whether it should be linear. What you're saying would be right if 4-win players were worth more than 2 2-win players, but as I showed in those articles, this is not the case. So, the expected number of wins from players is sufficient here. A high-prob low-ceiling guy can be worth as much as a low-prob high-ceiling guy as long as they are sufficiently high/low probabilities.

Thanks. I agree that most teams are going to victim's of the winner's curse. The solution to the winner's curse is to value the players at a lower level than you estimate them to be, figuring that if you do acquire them, they are probably not as valuable as you thought they were. My belief is that the D'backs didn't do that here, or at least not as much as they should have.

I guess it's possible splits could provide a clue about aging, though it would actually work the other way in your example, because a player could be platooned once they get below replacement level against same-handed pitching, so they'd lose less value in that sense.

Splits always need to be regressed back to the average split, because they are such a small sample size. I think that a good estimation is that you need 1000 PA on each side to even consider a player's true split skill level to be 50% his split and 50% the league average split. I'm not sure about the 1000 but it's something like that. Regardless, that doesn't matter much because Howard is better at RHP and worse at LHP. It doesn't seem to affect Howard's overall performance in high leverage PA, so it shouldn't make much difference.

You're right in a sense. I checked into this issue when I was preparing to create MORP in the offseason, and I found that going into the offseason, teams generally had replacement level players at about 4-6 starting spots, and almost always at least 2 spots. But that's as of November. Teams certainly have less "vacancies" now. However, most teams are still pretty close to replacement level in the 4th and 5th SP spots, which is why it's always easier to move a pitcher at the trade deadline than a position player. The Efficient Market Hypothesis economist in me wants to say that teams can trade away players in such a way to add to their roster, but these are high-friction markets, particularly in July, so that may be very unlikely. It's a good point that you make. Some teams are probably not going to be able to upgrade substantially even if it's an interesting question to consider how much they'd gain from an upgrade of 4.5 wins. Thanks for highlighting this.

I think that a big part of the huge boost in expected revenue upon making the playoffs is that there is a good chance even the worst team in the playoffs advances or even plays in the World Series. I'm not sure that the biggest payoff is the first round, per se, but I'm not sure. I guess there would be something more to be said for some of the teams with 20-30% chances of making the playoffs becoming buyers if I'm underestimating the gain from making but being eliminated the first round.

Adam Dunn is underrated and Ryan Howard is overrated. Hopefully, some team is smart enough to offer the Nationals a good haul on Dunn. I think that you're probably underestimating Dunn's relative OBP advantage and underestimating Howard's relative SLG advantage, but the reality is that their offense is very similar in value. The difference in skill is really that Ryan Howard has average defensive skill and the jury is still out on Dunn's defensive skill. Dunn has been better at first this year, but Howard's had a couple average years in a row at first since he lost weight and worked on his agility. As I wrote at the time, Ryan Howard's contract is going to depend on how well he ages and how much salary inflation baseball sees. He's worth the money now, but if he ages like Mo Vaughn instead of Jim Thome, that's going to be what determines if the contract is a disaster. If the economy sputters and salaries hold, though, it could be anyway.

I was basing things largely on Championship Odds Added, so the Red Sox at 4.0% were a little higher than the White Sox at 3.6%. In all honesty, I was updating this throughout the past week, and the White Sox number was growing and the Red Sox number was shrinking. If I did the analysis all in one shot, I probably would have called the White Sox a strong buy. Both are certainly up for debate. I still think the Twins should hold because of the low championship odds, or at least no more than a weak buy. They have a good franchise and a good window to win.

I assumed they were pretty good, but even if they are 5-10% off, the relative change in playoff odds added and championship odds added should be about right.

John, the Twins did have a huge jump in playoff odds added by bringing somebody in, but because their odds are not all that great in the first place, they have a relatively small gain in championship odds. I based most of my decision about what to do on championship odds. The Twins are a good enough franchise that they should be looking at that too, I think. Why sacrifice a championship later for a division title now? Elm, the frame of the article was that adding a player halfway through the season could have almost the same impact as adding him at the beginning. Except for circumstances like the Padres, the full season is going to help more than a little more than a third of the season in expectation. The point is that stars add about 2/3 the value for 1/3 the season for a lot of contenders.

My best guess is that the majority of selections were made before the last two weeks, and that sample is very unlikely to affect the 48 players selected. The performance of those players being LOWER in the last two weeks of the first half than in first two weeks of the second half would mean I am likely understating the improvement and fly ball spike if anything. If it helps, the last two weeks of the first half saw a 2.5% lower (FB+PU)% and a 2.6% lower HR/FB than the cumulative first half for participants, which presumably indicates that the drop in performance coming from regression to the mean was already occurring. My feeling was that the extra sample size far outweighed the concerns about the bias, though I agree a negative HR Derby effect would probably be flawed and I'd need to rethink the methodology; zero or positive effect seems unlikely to suffer from that issue to me.

Sometimes it is best to just look at odds of winning a baseball game. How much does having a solid starter increase your odds of winning a baseball game? Answer that question, compute the percentages, and then move the Secret Sauce stuff around at the margin to bump things up or down by a couple percentage points. Being better at baseball than one's opponents doesn't need a study to confirm it. Nate found 3 interesting things that shined above the noise, but that doesn't mean that the same things that make one team have a more than 50% chance of winning a game won't be important, too. You still need to assume the better team wins a baseball game more often than not.

Yup, it was a mistake on my part. It was higher post-Moneyball but modestly: .078 to .084.

Hmm...I'll find my spreadsheet tonight and report back. Thanks for finding this.

No, just saying that the difference between OBP and AVG for all-star starters was larger in this era. In the case of a .300 hitter, they would probably have smaller differences in OBP and AVG. Think of it like a .270 hitting power hitter generally had a .372 OBP in the Early Era and a .419 OBP in the Internt Era.

I'm looking into this, and it definitely might be worth writing a blog post about, so look for that. Thanks.

I would go for guys to make them competitive in a couple years if I were them. They don't seem likely to make up the gap in 2011 because they are giving up Fielder, but some prospects and some good moves could make them good in 2012 and 2013. It would probably involve trading some other pieces, probably.

Yeah, my feeling is this. The Brewers are probably a .500 team next year. Becoming a 91-win team would cost about $50MM beyond whatever raises players are already due in arbitration and already written contracts. Sure, they could lucky, but why risk it? That's not the point where you make a dent. A team that is maybe five wins from contention should drop some money on free agents, but I don't think the Brewers can cover the win gap with free agents without losing money in the process. It may be that I'm underrating them, but I don't have much faith in their non-Gallardo pitching.

My point was that all teams need young talent to produce. The Red Sox have done fabulously with young talent-- read through my "No Turnover Standings" articles. I'm curious why you think the Brewers are going to be competitive. Look at the Brewers' roster, contracts, and minor league prospects they have now. Sticking with Fielder, Braun, Gallardo, and the other talent they have is not going to make them a competitive team before 2011 in all likelihood (certainly not 2010), and retaining Fielder is going to cost enough money that it would difficult to do so afterwards. They are going to need young talent to get better-- replacing Fielder's uniqueness has a marketing argument, but replacing his wins is something that they should be attempting to do by trade.

The Red Sox spend $170MM, which would buy about 32 WARP, which would only put them at 72-90. The Yankees spend $215MM, which could buy 41 WARP, putting them at 81-81, meaning a minimal level of cost-controlled talent could push them over the edge. The Brewers probably should trade some other parts too, but players like Braun and Gallardo are locked down long enough that they are still assets in a couple years. Some other players would go, too, though.

It's a good question. In theory, I think it has to do with whether there is at least more seller than the number of buyers, at least one more buyer than the number of sellers, or an equal number. If there are 2 more buyers than sellers vs. if there are 3 more buyers than sellers, that should not change much in the bargaining (unless their replacement level is very different). The obvious comparable player potentially on the market is Adam Dunn. I would think three of the teams I mentioned might create the necessary scarcity to drive up the price. I like your question, though. I think the Rays might be buyers, though, at least according to reports.

There is certainly more roster flexibility in the offseason, so that's why you might not see the in-season bump for position players like you do for starting pitchers (there is always a spot for an ace). At the same time, the Brewers do not need empirical tests here, because they can simply start taking offers and create some urgency. If bidders do not realize the extra value Fielder provides, that would obviously be different, but I suspect teams do. I would like to look at this empirically-- I've found market inefficiencies before, and without savvy investors being able to make their own teams to compete in the MLB easily, you really can see market inefficiencies last in baseball unlike other businesses where competitors can emerge more easily.

So, what type of difference do you think the Brewers would have in attendance if they made the playoffs for a couple years in a row a couple years down the line. It seems like getting good sure spiked their attendance around 2007 and on, especially when they made the playoffs in 2008. Obviously this is going to mix in with the Fielder effect, but certainly the Brewers fanbase seems to respond to winning. Attendance is down this year even though payroll went up. I do agree that there probably is an effect of perceived effort by ownership, but a couple good prospects might make people think they are putting in the effort when they come to fruition even if it angers people now.

Definitely seems true about playoff hangovers, and I'm pretty sure they extend well beyond one year. Fielder's body type certainly isn't one that suggests rapid decline, but I think that would probably be priced into his contract at this point. Most teams with money are somewhat afraid of this, at least most teams without huge 1Bs.

Agreed that it's a big hit financially now. No debate that it would hurt. But I also watched the playoff chase and the 2008 NLDS and it seems like that fanbase is rabid for a winner, no? I suspect it would be a huge financial hit, but so would losing him to free agency. I just think a pennant could dwarf the lost revenue for the 2010-11. If they are going to retain him because he has such a unique value to them, that's one thing I don't have the hypotehtical data required to study, but I can't imagine a half decade of 80-win seasons is better than a couple of 75-win years before a couple playoff berths.

Thank you very much for pointing that out. I love Bill James and subscribe, but missed that article. The second game HFA spike is definitely ambiguous, and it's caused some contraversy before. There appears to be some psychological basis for it, as Russell Carleton tells me that sleep stuff supports that effect (you can tell by my diction who the expert is on this). Of course, this past decade provided stronger evidence than decades past. In SABR's Baseball Research Journal, I expanded this to look at 1970-2009 or something like that, and found a mildly significant effect. I'm inclined to think it's true, but I recognize smart people stand on both sides of the debate. Thank you for pointing out James' article on HFA. One of my favorite authors and one of my favorite topics, gotta love it.

It can have an effect for teams on the margin, unsure whether they want to trade veterans. There are a bunch of veterans practically guaranteed to be traded at the deadline every year, a million veterans who won't ever be traded at the deadline, and then a few guys on the margin. I think it can have an effect on the margin because the incentive is there, as I have mathematically highlighted above.

Yes, it's 13.18 wins out of the top pick and 2.32 out of the 30th pick using Sky's formula with the proper adjustments (rWAR to WARP, fraction of WARP out of first six seasons). I do think that it's subtle but I have to think that if there was a clear cut #1 pick again, the Orioles would be thinking long and hard about calling up a superstar prospect that might make up the difference between themselves and the Pirates, and I could really see the incentive for the Pirates to trade some veterans like Octavio Dotel just to do it. Once something like that happens, it's more dangerous. I'm not saying they are tanking yet, but when the incentive is clearly there, I don't see why the league should wait. Teams eventually do pick up on a lot of things like this, especially when there is a precedent in another sport or two.

The gigantic differences between SIERA and ERA are only going to come from luck. I mean, how big is a run park factor for a given stadium? Maybe, a hair more than half a run at the extremes? And defense probably can do something similar to that, I think, though I'd need to work out the numbers in more detail. The point is that if David Price has a 2.44 ERA and a 3.97 SIERA but plays in front of good defense in a pitch stadium, his defense- and park-adjustments to his ERA indicate he might regress to something like 3.5 instead of 3.97 but 2.44 isn't a realistic adjustment for park and defense unless you played in a ginormous stadium and with 8 superheroes playing defense for you.

The table has three columns. The first is the pick number, the second is the value added of the pick above the subsequent pick, and the third is the average record of that pick. So the average first pick is worth 3.90 wins more than the second pick, and the worst team wins 5 fewer games (58-104 vs. 63-99) than the second worst team. The average second pick is worth 1.73 wins more than the average third pick, and that team wins 1 more game (63-99 vs. 64-98). The average third pick is worth 1.04 wins more than the average 4th pick and that team wins two games fewer (64-98 vs. 66-96), and so on.

After playing with some numbers, I realized how arbitrary it would be to be specific. I've made some approximations of the breakdown of the marginal value of a win before (last July's Roy Halladay articles, based on some updates to earlier Nate Silver work), but I just didn't see a way to come up with a precise way to do it. But regardless of the way you do it, the win now is worth less than a large fraction of a win later, which is really what the table above shows-- that it's a larger fraction of a win that you gain for losing a game now. So the marginal value is going to create this incentive regardless of how exactly I break down the numbers. Play with some numbers, and you'll see that just about any realistic numbers you pick show the incentive for certainly the bottom four or five teams to lose, but that picking any ones on my part would have drawn away from the general point. I'm not talking about the Mariners trading away Cliff Lee for an elite prospect to help themselves. I'm talking about the Orioles trading away Kevin Millwood and Ty Wigginton because they are risking winning too many games by having them on board. It would certainly put a slightly higher price tag on them, because they would be more valuable to the Orioles. The arbitration argument is easily correctable-- a fixed number of years based on either age or draft year or signing year where you have them under control. If the Nats had Strasburg from the time he signed until eight years after they signed him regardless of when they called him up, they'd have called him up sooner?

I love this idea. I wonder if the league would ever implement it. Maybe you could have the same types of rules with playoff rosters where players need to be on your roster on August 31 to play in this series. That would certainly be interesting!

Right, you still want to lower those incentives. There are a number of ways to do it, but the current system is incentivizing teams to lose a little too much.

No, my point is that the lottery would lower the incentive to tank and that this incentive is starting to become more obvious.

Oh I agree with you. The ground that needs to be made up in evaluating defense is really large, and you don't want to conflate the two issues. The reality is that over a full season for a starting pitcher, BABIP variance is made up by 75% luck, 13% defense, and 12% pitcher skill. SIERA picks up on the 12% pitcher skill, and removes the other 88%. In half a season, even more of BABIP variance is luck. You can tweak SIERA by some small fraction of a run to come up with an approximate defensive effect just like you should make a park adjustment to add runs for pitchers pitching in hitters parks, but SIERA as defense- and park-neutral going to get you most of the way.

So your entire point is that since teams haven't been caught tanking yet (even though we see tons of talent traded away at the deadline every year), there is not a problem, and that since basketball teams still have some incentive to tank, you don't realize how much worse it would be if they had more incentive to tank. And you also seem to think that even though I have an actual table laid out for you showing that on average higher picks do better, you don't believe this is an issue because sometimes lower picks do better than higher picks and that you are unable to figure out the number one pick eleven months from now. There is less incentive to tank when the payoff to tanking is lower. If that is not evident to you, I don't what I can do to show it to you. If each game you lost now, gained you 0.8 games in the future when you might be competitive, that would be more of an incentive than if there was a lottery an that was lowered to an expected gain of 0.2 games. You also seem to think that since sabermetrics has already existed for many years in front offices, that means that they won't realize the value of higher draft picks in the future? Teams are learning things all the time. They're getting better at forecasting talent, too. This is about preempting teams from making decisions that appear to be in their best interest. Also, "you can win with the lowest payroll" is laughable. The correlation is high between winning percentage and payroll, and getting higher. Simply noting some counterexamples of smart teams does not change the fact that winning with a low payroll is hard. Having good talent evaluators helps too of course. That does not change the fact that having a higher draft pick helps in addition to having money and good talent evaluators. It's like you have decided that a few examples of other ways that teams have been successful changes anything about a backwards incentive structure. And you conclude with "if it ain't broke, don't fix it" when I conclude with being proactive rather than reactive. I like mine better. I've shown the incentive to lose, and now suggested a way to counteract it. That the NBA still has some problems with tanking does not change the fact that the race to the bottom is less valuable than before.

Maybe. It could be the other way though-- experienced pitcher is also older so cooling off may have worse effects. Tough to study but an interesting thought, thanks.

Meh, the Jays got a lot of potential talent in the Halladay deal. It ain't fun for this weekend, but the Jays time to shine is probably in a few years when Halladay costs a lot more than the talent they got for him and those guys are producing on the field.

The Milwaukee game was a neutral site. As I explained, home-field advantage comes largely from familiarity and distance. Neither team had any more familiarity with Milwaukee, both were staying in hotels, and there probably was not a huge difference with Chicago being closer to Milwaukee than Houston, but perhaps there is some. Regardless, I think the unfairness came from the fact that the Astros didn't get an advantage rather than any claim the Cubs may have gotten some.

It is absolutely ridiculous, agreed. I wouldn't negative rate you. But in all fairness, the interleague schedule is particularly unfair to the Phillies this year: http://www.thegoodphight.com/2010/6/25/1536636/for-the-phillies-the-injustice-of David S. Cohen of The Good Phight found that the Phillies had the hardest interleague opponents, with a winning percentage of .545, compared with the Braves .527 and the Mets .500. So the reality is that the .24 win bonus that the Phillies are getting here puts them somewhere between the Mets and Braves in terms of unfair unscheduling. I think it's pretty clear that it's unfair to the Blue Jays, and that a fairer solution might be neutral territory and actually making equitable interleague schedules in the first place.

Someone better come up with a way for the Phillies to relay signs better in their own park, because their home-road differential during the last 6 years is the smallest in the majors.

In general I agree that managers have incentives to play by a set of suboptimal rules to keep their jobs. However, walking the go ahead run into scoring position isn't really "by the book," is it? It seems like that would be the opposite situation. Also, Manuel probably has more job security than any manager you can imagine, so this might be less risky than other managers sticking their necks out on a strategic move.

There are a number of other likelihoods that I did not account for, all of which were small and had some other effects. The effect of a poor bunt only moved the probability a couple of percent if it happened, and a million other outcomes were also possible. Regardless of how you model it, you're going to come up with a situation where Danys Baez would need to be roughly halfway between a regular pitcher and a home run derby pitcher for it to make sense.

Of course there are a million little possibilities that factor in, with the single possibly ending up better than a walk or the double not scoring Mauer or whatever, but this was a rough sketch. They basically would need to think that this was almost a home run derby for Morneau to make this move credible.

If they IBB him and Rauch bunts, they have a 49% chance of winning. If he hits a single or regular walks, it's about 49% chance anyway. So you just need the odds of an RBI hit (which gives them a 13% chance) weighted by that probability by the odds of an out (which gives them a 64% chance) weighted by that probability to equal 49%. 0.49 = (p) * (0.13) + (1-p) * (0.64) p = 0.29 So a mix of a 29% chance of ending up with a 13% chance of winning and a 71% chance of ending up with a 64% chance of winning on average are as good as a 49% chance of winning.

I'm not sure you read the article very carefully. I looked at the BABIP in detail to determine how much of it was flukiness. Given the rate of ground balls reaching the outfield was flukishly low but the ground ball rate was not, given that the number of balls that reached the outfield in the air was the same as historically, I determined he wasn't seeing much of a change in his batted ball profile. I also discussed the normal range of BABIP that can be attributed to skill, which Ubaldo's numbers fall out of. Given that the drop in BABIP is a function of 40% of line drives being caught and 89% of ground balls being fielded by infielders, I think I have looked at it carefully enough like you would suggest.

Bob Gibson did not play in the era of batted ball statistics, so we do not have SIERA on him. I'm suspecting that his SIERA would exceed his ERA, but mostly because he record ERA was by definition lower than all his other ERAs and I assume that he must have had some good luck. Or, to put it another way, what are the odds that Gibson had bad luck that year? Really, really low, right? Therefore, I assume he probably had some good luck, especially because it was his best year. I doubt there are many pitchers that have ever had bad luck during their career year in terms of ERA.

SIERA is determined based on the current run-scoring environment that has been pretty similar over the last 17 years. That could change, at which point we wound change SIERA. The only difference in between OTSgamer's list and the true list is that the "P" term is Popouts, not putouts.

Of course, you're right that this is a limit. I was asked a similar question in the first No Turnover Standings article, and I responded with the following: "Playing time is not factored into it. There were 81*30 total wins/teams to go around, and I just added up WARP3's. If you want to think of it as the production of the amateur scouting department, with an idea that players could be traded at fair value but the No Turnover Standings are a reference point, that works too."

I think it's just a matter of me explaining the method more clearly. The Braves record of 89-73 means that they were 8 wins per year above average when you add everything up-- 4 above average in the 1990s draftees, 6 above average in the 2000s draftees, 1 below average in the 2000s signees, and 1 below average in the 2000s Asian professional, and average at everything else. I'm sorry I wasn't clearer, but it should be that way for every time, give or take rounding adjustments.

Of course. I definitely want to look a little more into draft pick number, because I'm sure that's driving a lot of this. It's not really comparing apples to oranges in the sense that I'm still asking what teams produced and getting a good answer, but obviously some teams have advantages in terms of the draft. This just shows the resulting differences in acquiring amateur talent.

Yup. Please see the last article for details. The Rays actually do an incredible job of acquiring other people's minor league talent. Think Ben Zobrist, J.P. Howell, Matt Garza, Jason Bartlett-- all these players were acquired in trades the Rays won.

I would look at SIERA over the past three years, obviously weighting right now a lot more to get a defense-neutral version of ERA and then just adjust by time. Look at how the teams FIP compares to its ERA, and figure that the difference between SIERA and ERA should be similar to the difference of the team FIP and the team ERA (though maybe shrink this difference a little bit). The fewer innings you are looking at, the more you want to regress the whole ERA back to the mean, though. For fantasy baseball, I would imagine it is probably more important to rank pitchers ordinally rather than giving a cardinal measure of ERA though? In that case, SIERAs with a FIP/ERA adjustment might do the trick pretty well. Comparing SIERA and PECOTA can probably do a good job of giving you extra information too.

Ah-- yes, league average defense.

Well, it tries to predict ERA going forward as much as FIP does, only it incorporates a few other factors to make it less biased against certain pitchers. It tells you what ERA typically is for pitchers with certain GB, BB, and K rates, taking into account the interactions, etc. It also incorporates BABIP skill that pitchers do have. So FIP would assume that pitchers have the same distribution of BIP skill, while SIERA assumes that pitchers with high K% and low GB% will have lower BABIPs.

They are in the four reports-- the Pitcher Standard Report, the Pitcher Rates Report, Pitcher Team-Standard Report, and the Pitcher Team-Rates Report. You can also get SIERAs if you do a customized stats report. I have a couple links in my favorites because I like certain things all together in one report that I can make a .csv out of.

Certainly interesting, but not really enough to explain things. We do see a spike in HR/FB for the D'backs offensively from 10.4 to 11.9%. However, the spike defensively is 13.0 to 18.7%, which is just way larger. If it were only year-to-year changes in park effects, that jump could not be as large. Good point, though.

This is true, but I think most of what the article is doing is actually explaining this fact, rather than calling everything luck. That said, there IS such a thing as luck, and you CAN actually pinpoint it. There is a counterrevolution in sabermetrics claiming that too much is being called "luck", which is a good thing to highlight, but it's being taken too far. We KNOW that a binomial variable like BABIP should have a standard deviation of .019 for a whole season among pitchers who throw 150 IP. That means that when we observe an actual standard deviation, net of team, of .021, we CAN see that MOST of BABIP fluctuation is bad luck. Of course it's important to explain all the variance you can, but dismissive statements saying that you can't call anything luck are as dogmatic as saying that everything you can't explain is luck.

So that's a really small effect, and also somewhat in the wrong direction. Dan Haren has surrendered 5 of his 16 HR in the 1st inning, and 8 in the first three innings. Edwin Jackson has surrednered 5 of his 10 HR in the first three innings, while Kennedy and Rodrigo Lopez each have 5 of 11 in the first three innings. We have a stat called Fair Run Average (FRA) which takes into account what RA should be adjusting for inherited runners. Haren is at 5.16 RA with 5.22 FRA, Jackson has a 6.03 RA with a 5.96 FRA, Kennedy has a 3.50 RA with a 3.35 FRA, and Lopez has a 5.20 RA with a 5.07 FRA. These type of things are small effects compared to the larger issues cited above relating the BABIP, HR/FB, and difference in performance with men on.

I mean, the t-test definitely makes more sense to run if the data has a unimodal nonskewed distribution, because that will at least look normal enough to make a t-test relevant. The thing is that the data is so incredibly different that really any t-test will either tell you "there is enough data assuming that the samples aren't biased" or "there is not enough data". The only thing a t-test can tell you is if you're ready for a t-test, because it would be a problem that the sample size was too small if the t-test failed to reject the null. Failed to reject the null is all you could say...certainly not enough to say you accept the null. $/WARP or WARP/$ seems like a judgment call but it would probably be based on what the data was shaped like. I guess looking at changes in total WARP vs. % WARP change is useful too, but it won't grab the injury effect for mediocre players (because going from a 1.0 WARP to a 0.0 WARP injured player isn't that much of a change, but going from a 5.0 WARP to a 2.5 WARP player will seem much worse). I think that could actually be a good way to look at things, but I just think it's tricky with playing time. Maybe absolute changes could be good for looking at rate stats? I'd need to think about that. Absolute changes in counting stats relative to a replacement level just could be really biased, especially for two-year deals often given to players not that far above replacement level. You did look at three-year deals at least though. I see a little more what you're doing now, but be very careful not to read too much into t-tests. The real question is (a) if we have biased samples, rather than (b) if we have enough data. The "is the difference large enough" question is obviously not legitimate. The question is really about the sample size being large enough to confirm the difference matters, because the difference in performance is huge.

How can you run t-tests?? What did you do for a standard deviation?? I don't think sample standard deviation makes any sense at all in this case, and I don't think you can make the necessary assumptions about normality or whatever else would be required to even assume a sample standard deviation works. Also, the means for the three-year deals are very different so any changes in expected average wins is particularly useless as a measure. The third year new signings lost half of their value...how can you compare that to the re-signings who lost only 20% of their value? I don't see how you can run t-tests really at all on that. Nextly, there are 33 new signings of 3-year deals, not 26. So right there I'm not sure what you did either. Also, where are you getting 0.9 WAR/(Salary/3)? Maybe 0.09? Is that what you meant? Why did you use wins per dollar instead of dollars per win? I can't imagine wins per dollar had a normal distribution either, so you're really confusing a number of things I think. The three-year re-signed had a 3.6 WARP3/salary in the third year, and the newly signed had a 13.9 WARP3/salary in the third year. That's different by nearly 400%. If whatever test you're running doesn't find that significant, you're not running a useful test. It's possible that whatever tests you're trying to run, even with the standard deviations that you could even use to make sure, just require a larger sample size. Someone on on another website checked my thesis for 1990-2009 and found that the results were on a similar scale by slightly smaller for this period of time. So if your post is a way of saying "we need more data," that might be a good place to start.

In all fairness, his late development was because A.J. Burnett broke his wrist, and he needed a couple surgeries to fix it and he got DFA'd. He was a top prospect before that. Also, he plays a corner but he's capable of playing center. He's very athletic, and probably doesn't fit the "corner OF uh oh" profile. I don't think there's any reason to assume he'd be overpaid as a FA. It seems like he's the type of player who can push them to 90 wins in 2011 at least.

The Phillies raised payroll after Cliff Lee left. I think they traded Cliff Lee to restock the farm system. They had in mind that they were going to increase payroll by about 10%, and they thought they could restock the farm by increasing it other ways. Keep in mind that Benny Looper, the Phillies' Asst GM, was the Asst GM of the Mariners when those three prospects were brought into the Mariners' system. I think the Phillies specifically wanted those guys. Regardless, they raised payroll by about 10% and are likely to keep raising payroll by about that much, I think.

AND Olmedo Saenz too. He wasn't eligible for free agency when he re-signed either. Again, doesn't change the results but worth mentioning.

NOTE: Jason LaRue shouldn't be there in the two-year deals. Just realized he was a few days short of free agency when he signed the two-year deal in the off-season of 2005. Doesn't really change much analysis wise, but should be change in the dataset.

I have a hard time believing the hometown discounts are 50%, especially because the real difference in the $/win performance from these players is a result of performance late in the deal, not early. If it were only hometown discounts, you would expect the WARP3 values to decline at the same pace for both re-signed players and OPP, but instead you see the real production different late in the deal. That smells like something far more than a half-off discount given to the hometown team.

Very good point. I got rid of contracts that covered arbitration years in this round of analysis to make sure that the risk-aversion of that sort wasn't mixing in with the results, but I actually think that there could be an element of exactly what you're talking about. Players don't want to move their families, etc. On the other hand, that probably isn't the whole issue, because otherwise pitchers must be far more risk-averse individuals than hitters are!

Interesting. I didn't think about the four-year deals being biased that much, but the breakdown of pitchers and non-pitchers is perfect. It's the same result after all! I think this only says that pitchers are overpaid if the replacement level is perfect for both. If the replacement level for pitchers is overestimated or if the replacement level for hitters is underestimated, it may be fine. One thing that this makes me think is that maybe there is a rational basis for teams over-drafting pitchers. They seem to be such bad bets based on Sky Andrecheck's draft pick analysis, consistently producing less than hitters drafted in similar rounds. Hitters on the free agent market being better bets than pitchers might make reevaluate this. I guess it would depend on vacancy availability to sign hitters. i.e. how many lineup spots are replacement level for a given team vs. how many rotation spots are?

The players who signed deals with the hometown clubs still may not have given hometown discounts. Certainly this magnitude of hometown discount would appear unlikely. Economically, the lack of testing the market shouldn't be an issue on its own, because the player and team should both have a good approximation of how high bidding wars go, and the player should want his hometown team to match this approximation. It seems unlikely that player agents would perpetually undervalue the effect of a bidding war. If they expected a bidding war, they would typically force the hometown team to bid up for it. Really, I think this comes down to information asymmetry. That's the "why" in my opinion at least. The evidence that pitchers show a more extreme effect probably adds to this case.

I definitely agree the opportunity cost of re-signing teams is larger because of the supplemental sandwich pick, which effectively costs them 2 picks for signing a player instead of 1. I'm not sure that would explain the difference though, given the magnitude. The extra pick is probably worth about $6MM per contract. Also, if there were perfect competition between teams, the hometown teams would simply be outbid because of the extra "tax" of the sandwich pick.

I agree a longer time period is a good idea. I don't have service time information that goes back all that far, so it makes it somewhat difficult to discern which extensions cover arbitration years and which extensions cover free-agency eligible years. If I could get my hands on this data, that would be great. It took a long, long time to aggregate all this data already, though. I agree that more would be ideal to see this effect through time.

I like the idea to look into hitters and pitchers separately! I forgot about that. I had started with a few splits into this issue in the first article on this topic a couple months ago, but I should have checked it again when it came to money. Thanks for breaking it down. I do agree that Howard isn't the best example of a player who is likely to age differently than others. I think the only reason that he would be an example of the "information asymmetry" issue is because of the supposed workout plan. However, Howard was the big extension contract du jour a couple weeks ago, and I mentioned it in the article as a reason. I hate hitching my analyst wagon to the "Howard deal is good" simply because I said "it might not really be that bad." I'm still not thrilled about it as a fan of the team. Still, the comment about not needing any tie-in to Howard isn't really relevant. The Howard article was a great way to re-broadcast this research which got burried without much comment until Howard was signed. I'm a practical guy, and I knew that had a good strong enough relationship to my result that it was worth re-stating clearly. Chipper, on the other hand, is an injury-ridden superstar. Clearly that makes him a better candidate to be a bargain based on information asymmetry. This is all the more relevant because it's the Braves, who are by all reports a tremendous scouting team who does the due diligence on these type of issues. If the Braves think Chipper's unhealthy after his deal is up and don't re-sign him, I sure don't want him.

Tango, it appears you've selected on the dependent variable. Put it this way-- if you remove the OPP guys who failed to produce after the deal was signed, you're eliminating a lot of data that could be the exact evidence of what we are looking for. Why is the fact that Adam Eaton and Jason Schmidt were disastrous three-year deals a lack of evidence that the Rangers and Giants knew something that the Phillies and Dodgers didn't get???? The point here is that the money is a proxy for how good they were before the deal. Eventually, I can dig up the WARP in previous years (if anybody has added to the data sets I provided above, please let me know), but the money they were signed for is the real value. I'm in agreement that this work isn't done, but the cost per win is really strong evidence that these groups are dissimilar for the reasons mentioned. What we have here is evidence that it cost more to get wins from OPP than your own re-signed guys. Regardless of how good or bad the players were beforehand, the players allowed to reach public bidding were real bad contracts on average.

Mariano Rivera averaged being worth about $9MM per year from 2002-09. That just seems low to me. I can't breakdown WARP by relievers and starters, I don't think. Maybe there's a way to do it in the sortable stats. The relievers' salaries might be misleading. What's the average age of a reliever vs. a starter adjusted for leverage of innings? This could make the cumulative salary vs. cumulative wins comparison biased if different. I'm not saying either methodology is right for relievers. I'm just saying it's a hard topic, and probably one where there are improvements to be made.

I don't like the use of FIP for relievers. I think it's inconsistent to attribute the BABIP luck that hitters get completely to themselves, but to leave it out for pitchers. That's really my issue. The chaining thing I'm not sold on yet, but I'm not sure it's wrong either. I'd need to look at usage patterns in a lot of detail. Colin's criticism about the creation of leverage seems concerning. It also seems like replacement level relievers get so many high leverage innings that I'm not sure that the absence of a player would always bump up the leverage of all other players one by one. It also seems a little bit inconsistent with the concept of replacement level, but maybe that is resolvable. I'm not saying that I like it or don't like it, just that the results of such low win values give me pause. The WPA values are so high for good closers, and it seems like chaining is the reason that the fWAR values aren't. It's just something to think about harder, and an outcome that says FanGraphs thinks 30 teams overpay for relief help even though WPA seems to imply that they might be fairly priced...it's just something I'd want to devote a lot of time to before forming an opinion.

Little short-- he had 1.3 WARP3, which gives him a MORP of $6.8 million, so his $14.3 million cost, leaves him at -$7.5 million value, which fell short of the list.

My understanding is that they are dealt with in a similar way as starters. The player's Run Average is adjusted for defense, park, and opponents, and compared with what replacement level players would do with similar defense, park, and opponents. Runs are then converted into wins. WXRL is done differently as I understand it, but I kept WARP3 for relievers to compare apples to apples. Approximating win values for relievers is really tricky. I'm not sure that there is a great way out there to do. FanGraphs uses chaining and FIP, but the win values come out very low, and there are a lot of assumptions built into that which I'm not totally comfortable with.

Yes. I approximated signing bonuses by draft slot. It doesn't change the numbers much, but it does play a role.

Okay, it seems like giving out my spreadsheet won't be an issue. I don't have all the data you ask for but I do have names, ages, deal lengths, and WARPs. I'll try to put something together soon, maybe a google doc.

The majority of this was from a proprietary data set that somebody gave me, and a lot of it was pieced together from that and other sources. I'm not sure I'm even allowed to give it away like that.

I agree the 26-31 two-years are a useless sample in isolation actually. I should have said something about that, but I forgot when I was forming the tables. I decided to include them because they are part of the whole sample and I didn't want to be removing players from the sample. But Andruw Jones -2.0 WARP the first year of his Dodgers' deal is such a matzoh ball hanging out there that any way I did the ages, some group was going to be stuck with 31-year olds and they were going to age badly. And Abe Nunez was pretty ugly at 30 years old with -1.6 WARP the first of his two years w/ the Phillies too. In future articles, I'm going to make a point to avoid using percentages alone. I've gone back and forth using percentages and total WARPs, but I think I should use both.

In this article: http://www.baseballprospectus.com/article.php?articleid=10505, I looked for this. I got all the PECOTA projections for each group of players. The re-signed players underperformed their PECOTAs by 13% and the newly signed players underperformed their PECOTAs by 15%. That's basically the same. Sean Smith did that study without doing it by subgroup that showed CHONE over-projected free agents by about 15% too. So it seems like they are performing similarly on aggregate the first year. I think that's selection bias-- I discuss this in that article. I agree that I don't have perfect matched pairs. I think that's why it's important to aggregate to get some sense of these things.

Thanks to all those asking for WARP by year totals. I had forgotten about this. I'll do them in the comments, but I can post an unfiltered post if it's not clear. AVERAGE WARP EACH YEAR FOR EACH DATA SET (N = SAMPLE SIZE IN PARENTHESIS): TWO YEARS 26-31 YEARS OLD Re-signed: 1.64, 2.48 (N=9) Newly signed: -0.34, -0.18 (N=5) 32-34 YEARS OLD Re-signed: 0.37, 0.53 (N=12) Newly signed: 0.28, 0.33 (N=11) 35-37 YEARS OLD Re-signed: 0.80, 1.31 (N=13) Newly signed: 0.73, 0.44 (N=7) 38-47 YEARS OLD Re-signed: 1.00, 1.80 (N=5) Newly signed: 1.16, 0.44 (N=9) THREE YEARS DEALS 27-30 YEARS OLD Re-signed: 1.99, 2.11, 1.59 (N=9) Newly signed: 1.13, 1.43, 0.73 (N=12) 31-34 YEARS OLD Re-signed: 4.13, 3.73, 3.68 (N=4) Newly signed: 1.17, 1.35, 0.29 (N=15) 35-38 YEARS OLD Re-signed: 0.40, 0.31, 0.29 (N=2) Newly signed: 1.96, 0.49, 0.48 (N=3) FOUR YEAR DEALS 27-30 YEARS OLD Re-signed: 1.47, 4.17, 2.87, 1.03 (N=3) Newly signed: 1.50, 1.85, 1.77, 1.57 (N=6) 31-34 YEARS OLD Re-signed: 2.27, 2.03, 2.53, 2.17 (N=3) Newly signed: 3.89, 2.87, 2.23, 2.86 (N=7) So, all in all, it doesn't seem like there is any real pattern to the overall quality of the players signed. The sample sizes obviously get pretty small, but on the aggregate, they were statistically significant, and nearly every one of the individual samples seems to point in the same direction. I'll continue to look for different ways to splice this data set and please make suggestions if you have any. Thanks for these comments.

There is also a negative linear term in there for regular strikeouts that will always dominate. You would need to have over 100% strikeouts to have the positive squared term have a bigger effect than the negative linear term. Think of it this way: ERAs can't go below 0. The positive squared term slows the run prevention down when there weren't many runners on in the first place.

Oops-- the $24.8 is right, but I typed in $34.1 for MORP instead of $38.1. The formula for Doc would be ($4.9MM/WARP)*(7.7 WARP) + league min 400K = $38.1.

Actually-- it wasn't $1.88 billion for free agents. It was $1.69 billion for 383 wins, plus the draft pick compensation, which made it $1.63 billion for 337 wins (subtracted out bonuses and the discounted win values of the draft picks last). That would leave about $810MM for arb-level players about, so that would be almost exactly $2.0MM/win for them.

Blake was acquired in July 2008 at the end of his contract with the Indians. He was granted free agency after the year, but re-signed with the Dodgers. I agree the trade market could shine some light on how players are valued, but that is not something that could be modeled in the free agent market. I've seen some models of trade valuation, but there is room for improvement. Mid-season trades have different methods of valuation because MORP isn't linear through the year. Wins are more valuable at the beginning to teams who don't find themselves at pennant races in the end, and more valuable at the end to teams who find themselves in pennant races in the end. If teams knew how close their races would be, MORP would be uniform through the season, but since they don't, it's not. Trade value gets really complicated for that reason.

I doubt he's doctoring the ball. I'm pretty sure that he could make the batters miss sometimes he did-- he gets swinging strikes at about half the league pace. The ball just seems to be going right at the Nats' fielders no matter how well hitters square it up.

Well, that's mostly because he got paid $14MM less. Rivera had quite a whole to cover there. I think Rivera outperformed his contract last year, but not as much as Counsell. The reason Counsell comes out looking so good is that he was a borderline replacement level player who suddenly had his SLG jump 100 points while his AVG spiked 60 points. When you get an above average middle infielder for free, that's really good. Rivera was paid like an elite closer and performed like a super-elite closer.

I really like this idea. I'm going to see if I can put something together. If it turns out that it's just a bunch of obvious names, maybe I'll make it an unfiltered post or something. Thanks!

Ah, I can clarify then-- The salary cap would be to make players affordable for small-market teams. Small market teams clearly have an avenue to cheap talent already, though.

The WARP3 numbers are somewhat of a methodological difference from fWAR. WARP3 has a lower replacement level than fWAR (~1250 WARP3 vs. 1000 fWAR, 49-win vs. 40-win replacement level). That's probably a judgment call in that range. I'm not sure how big the replacement level error bars should be. As far as the free agents, I don't have my numbers in front of me but I'm pretty sure it was 383 WARP3 for about $1.88 billion, but that could be +/- .03 billion. The net (MORP-Salary) sum total was 0 by definition. Using my service-time contracts and wins articles, I can see it was about 403 WARP3 for arb-level guys, and min-salary guys had 445 WARP3. I think the total salary compensation for the league was something like $2.8 billion, which would be $2.5 billion above the league minimum. So it's about $630MM for arb-level players above the league minimum; so about $1.6MM/win. There are three main reasons MORP might seem high. 1) Draft pick compensation-- it was 12% of all free agent compensation in 2009, I think; 16% and 15% for '07 and '08, if I remember correctly. That number is based on discounting future wins at 8% per year, so it could be higher or lower depending on that. 2) Using all years of a contract, rather than just the first years. The market moves throughout the off-season and before. No one knows the exact marginal revenue, but they try to pin it years in advance, not just the off-season in advance. So when a two-year contract is signed in Feb '08, it matters for average salaries in '09 even if it wasn't in the 5-month window before '09. I had an analogy to BLS employment numbers in my previous article, and how you can't look at wages only of new job-finders to find average wage levels. 3) Using production vs. projection. Projected free agent production is about 15% higher than actual production, according to PECOTA and CHONE (CHONE is according to Sean Smith/Rally, I only did the PECOTA test and confirmed the same bias). I had an article a couple weeks ago about how players who reach free agency are a biased sample of all players so they shouldn't really be up to par with comptuer projections that don't know the information contained in the economic decisions of their teams.

No, not at all. A lesson from this would be that the National League teams have been able to avoid paying for better talent in the draft by mimicking each other's behavior, and have managed to win their share of World Series anyway. Unless it is hurting their bottom line that fans in their cities know the American League competition is superior, the lesson very well might be that teams could do well to simply draft only the more affordable players and only change that strategy if teams in their division were doing anything different. If paying more for these draftees only encourages your rivals to do the same, it might not be beneficial to do so. Of course, if the talent level got so much worse that NL teams could not win World Series, that would be a different story.

Thanks. I found Steve's articles. Very interesting stuff-- the difference in methodology (other than WARP vs. WSAB) is that he looks at what minor league systems developed the players and I looked at who drafted the players. I guess this somewhat reflects something about my thinking at the time that I had not realized-- I was basically looking for who identified talent well enough as a springboard for who helped themselves the most from that starting point, while Steve was clearly asking who was capable of developing talent, rather than identifying it. I wouldn't have expected the results to come out so different, but they really did for 2007, which was the only season of overlap (he did 1946-2007, I only did 2007-2009). The teams are more or less in the same places, but with about half a dozen teams swapped in different orders. The NL East totally changed shape compared to what I did. Interestingly, the Mariners were not the dominant superpower in "value production" as they were with "no turnover." There must have been a ton of players who were drafted by the Mariners by traded away before being developed. This could be because the Mariners aren't given full credit for David Ortiz in Steve's work, and Ichiro clearly doesn't count based on his methodology. I'm guessing there are other players I can't dig up right now that might be causing this big difference. It's interesting to look at the differences in outcomes for a subtle change in what question is being asked. Thanks for pointing out this article to me.

The claims that the economy is getting worse are not really accurate. The economy is growing again (relative to where it was several months ago). Unemployment lags the economy by about six months typically. Firms don't start making hiring decisions right away. So, much of the awful labor market is a result of how bad the economy still was six months ago. It's not a consolation for anyone out of work, but it does mean that the labor market should improve. Nextly, there are a lot of jobs to make up. Unemployment is high in large part because of the cumulative amount of jobs that have been lost. The other thing is that March finally saw significant job gains, which is good, but also very well might make the unemployment "rate" look artifically high as people who were not even looking for work now start looking, which makes them count as unemployed instead of out of the labor market altogether. The recession is over-- meaning that the economy stopped getting worse, in the sense that the economy is growing, but also in the sense that it actually has a lot of catching up to do and in the sense that there is a major lag in labor markets, a lot of people don't feel it yet. Personal consumption expenditures actually don't lag the economy like the labor market does, and personal consumption expenditures are increasing in recent months and are higher than last year. This would certainly indicate the demand for baseball tickets might be increasing right now. That does not mean April attendance would rise-- something Neil explains well a few comments up the page-- but it should indicate future attendance would go up. The problem, of course, is that attendance is really just quantity of tickets purchased. As a number of readers have pointed out above, the question of prices is another matter, and teams may be raising prices enough that the revenue may be higher even as the attendance might be lower. What might actually be a major issue is something Shawn has written about a few times recently-- the strength of substitutes for watching a baseball game in the stadium are various forms of media. If those substitutes are becoming superior, this makes the demand for tickets go down as the demand for baseball goes up.

The slotting undoubtedly makes it easier to coordinate, but it is perfectly reasonable that coordination could occur without any signalling by the league. Teams could simply observe each other's behavior and react accordingly. That is exactly how teams were able to suppress salaries during free agency when players also could not select their teams. In fact, it does not seem like a coincidence that this type of coordination between teams would occur around the time that free agent salaries really began to escalate. That would probably strengthen the case if I had to guess. Thanks for the comment. It gives me a much fuller understanding of the situation. I'll definitely look for Treder's work on this topic.

That could be the case. It could also be that they simply find it worthwhile to outbid NL teams for some free agents and are willing to pay more. I would guess that they are just slightly less efficient in recent years by chance, but it could be more complicated than that.

I will try to track down this research. Thank you for pointing it out to me.

Thanks! I really appreciate this, and I'm glad you enjoyed the article so much.

A big part of the question is whether those guys would be available at the same price. I don't think they will, but I'm guessing that the price difference would be worth the difference in production. I could be underestimating salary growth, like many of us did during the last economic expansion.

You said that "the Phillies are competent" was not a given. That's almost calling him incompetent, at least. I guess you're probably right that it wasn't exactly the same comment, but it's how I read. I also didn't know who Abner Doon was, to be honest, but a quick google search seems to imply he's not a bright guy, right? Regardless, I think we agree that Amaro can't be given the benefit of the doubt automatically, but perhaps we also could agree that it's worth considering he is smarter than his moves make him seem at first blush. You are correct that trading Lee was not part of trading Halladay. They just happened concurrently. Polanco was already signed at that point, so really I meant that they could have just non-tendered Joe Blanton and Chad Durbin, and not signed Baez and Contreras. Those guys cost about $14MM this year. My point was just that was totally okay. I am also skeptical that he necessarily knew what he was doing with Ibanez, but it certainly seems plausible that he did. I also agree that Amaro inherited a stockpile of resources, but keep in mind that he was seemingly a very involved Assistant GM who may be responsible for such good fortune during the Gillick years. Again, just asking for some hesitation at blasting Amaro. I've gotten burned on doing so a few times.

Agreed. I'm not saying that they were actually good moves given the information he had. I said it's certainly plausible. Ibanez isn't as slow as other plays who are 15 runs below average at defense, and the Phillies seem to be very good at positioning their defense well. I suspect that it was a reach that he would improve 15 runs, and we'll probably see a regression, but it's possible that they thought he had coachable flaws given how poor his routes or positioning was or something like that. However, I do think they knew Burrell was going to tank. He had injury problems that he played through for the last several years. I think that they sheer wrongness of my blasting the decision to offer Burrell arbitration was rather humbling for me. However, I agree with you the jury is still out and there are major question marks surrounding many of the moves.

I'm as skeptical as the next phan, but this statement is just ridiculously unfounded. His major moves include not offering Burrell arbitration which led to a massive attack by a million analysts, myself among the loudest, after which Burrell got WAY less than he would have in arbitration and played replacement level after words. He signed Ibanez to replace him, a move that was lampooned, right before Ibanez's defense appears to have improved by like 15 runs and during which Ibanez produced about 2/3 of his contract value during the first year of the deal. A disasterous 2nd and 3rd year would make the contract still look justifiable ex post. He signed Moyer, which seems bad but not disasterous considering he's only really paying him to provide 1-1.5 wins. He brought in Chan Ho Park when no one else would and squeezed a ridiculously good year of relief out of him (albeit after a failing attempt to use him as a starter). He also signed a large number of players to multi-year deals during the 2008-09 offseason, well below what similar players were getting paid, likely taking advantage of the good will towards the team that winning a WS will do. Then he cut ties with sunk costs like Geoff Jenkins. In the middle of the year, he leveraged the Jays and Indians until one of them gave him their ace at far below the typical trade value for players of that stature. This past offseason, he signed Polanco (an average player) to a contract that you would typically pay a below average player, signed Blanton to a three-year below his market value, and made a contraversial trade of Lee and prospects for Halladay and prospects that looks questionable, but keep in mind that we haven't actually seen what happens with those prospects yet. Then he signed Halladay to a below market extension. Does he do everything right? No way. But to call his moves incompetent or idiotic, or to say he's made far more poor moves than good ones, is just untrue. The jury is still out, and he's proven me wrong in a couple times already in my skepticism. I have an open mind for sure.

Great questions. Thank you for these. I think that I have taken these into account implicitly, but they hit at the heart of the economics of baseball. I'll do my best to answer them well, but please ask me to clarify if I am unclear. The marginal revenue of a win is most for teams on the playoff bubble. Of course, there is no way to know exactly how many games you and your division rivals will win. One win for the 2007 Mets would be worth tens of million of dollars in expectation, and one win for the 2007 Phillies would be worth only a few hundred thousand in all likelihood. Since the free agent market is actually an auction, the prices reflect the second highest bidder's valuation of the player's services. Mostly, this would mean that they reflect the value of a player's services to a team on the bubble. This actually means that 1 win is worth itself in the standings (ignoring playoff implications) PLUS a 5% increase in the probability of making the playoffs. A 4-win player represents four wins in the standings plus about a 20% increase in the probability of making the playoffs. Thus, this is pretty much exactly what is being paid for by teams. So the price of a win is basically a competitive team's estimates of the long-term revenue implications of one win in the standings and 5% higher playoff odds (and naturally, all the variance in possible outcomes upon making the playoffs). Thus, Howard's contract takes into account the possibility that if he takes a Phillies team that would otherwise be three games back in the division and makes them win their division by one and that they happen to win the World Series that year, then he will have been worth like $100MM (to throw out a number) to them. It takes into account the possibility that he takes a team that would have won the division by one games and makes them win by five games in a year that they lose in the first round, and thus makes the team $1MM or $2MM. It takes into account all of these things. The only thing it really assumes is that the Phillies are pretty much competitive going into each season for the contract. Now, if the Phillies turn into a bad team, the contract will be a bad one if they are stuck with him. Of course, they can trade Howard at that point. They would likely have to eat money, but that's fine, because back-loaded deals effectively entail paying for early-years' performance later on, so eating money in a trade would be exactly that. It also takes into account the variety of ways that Howard's performance may change so that if he turns into the best player in baseball or replacement level, trades in all scenarios are priced in.

Well, he was worth about 4.5 WARP on average the last three years. For the sake of picking numbers, I projected he'll start the deal a 4.0-win player and end it a 2.8-win player, so maybe 17 wins over that time period. If I had to guess, he'll probably end up being worth only $110MM over that period after taking into account the two picks he cost the team. I would put huge error bars on all of that, so please keep in mind that I'm just picking for fun.

Firstly, Adrian Gonzalez is only 2.5 years younger than Howard. Secondly, I'd bet he gets more than $25MM/year. He would get $25MM/year if he went on the market right now, but I'm guessing he breaks $30MM/year. Both teams gave up leverage. The myth that future price uncertainty isn't already included in current prices continues.

That's a pretty bad mis-characterization of my Werth section. As I said, the reason I believe it's not a tradeoff is that they are more likely to make money if they spend at the playoff bubble rather than intentionally falling short of it. As to the Cliff Lee claim, that was clearly not just to save money. They spent far more than his $9MM in salary in subsequent moves after the trade. That trade was about the value they placed on Seattle's prospects, guys that their own Assistant General Manager drafted when he was Assistant General Manager of the Mariners. The Phillies decided on a budget based on what they felt gave them a competitive team, and they chose to trade Lee and spend the money elsewhere because they thought the prospects were worth the difference in between Lee's production and the combination of Blanton's, Durbin's, Contreras' and Baez's production. As I wrote at the time, that was a questionable move, but subsequent actions clearly revealed that it was not just to save the $9MM. Otherwise they wouldn't have spent more than that weeks later.

I can't remember who wrote what. I just remember a lot of comments questioning the timing and asking how well second basemen age. Sorry.

Every time a new huge and long deal comes out, people look at the contract and assume there will be little inflation and determine it's a disaster. With the exception of the last couple years, this is overwhelmingly a mistake when it is done. Assuming a lack of salary growth in a rapidly expanding industry seems far more likely to be a presumptuous assumption. I hardly think it's fair to say I'm taking MORP too seriously or misapplying it considering I only introduced in five days ago. I have a pretty good sense of what it represents, which is the opportunity cost of a win on average. That seems to me like an excellent way of assessing whether a deal was smart or not. This article took a lot of time to write, and hardly represents some sort of contrarian cooking the books. I have been hammering the point about the differing aging curves when you compare re-signed players and newly signed players for weeks. It was my exact response to the Mauer signing as well, which also represents a player where there is a lot of inside information that the team has on the player's health. What I tried to do was look for what kinds of scenarios would make the deal worthwhile and what kinds of scenarios would not. The majority of the sabermetric community appears uninterested in considering the variance in performance and the variance in inflation to simply arrive at a sure answer. Instead, I attempted to look at a range of possibilities objectively without the dogmatic claim to predict the future with certainty simply because I have a weighted mean. I looked at the distribution instead of the weighted mean, and discovered that the contract was probably not a good one, but hardly the disaster that sloppy analysis suggests.

The .261/.348/.529 is (I think) in games that were started by LHP, since Baseball-Reference.com has 3232 PA combined in "vs. LH starter" and "vs. RH starter", which is Howard's total PA. So that line appears to include him mashing RHRP after the LHSP had departed. I guess the implication is that you should play him in those games anyway, because you'll have used up a bench player in the 6th or 7th inning when you replace the RHB who started against the LHSP.

Oops, I responded to this post above as part of my response to the other post. But the point I made is that it's not a perfect correlation but the oppo-mashers on that chart did age better than the pull-mashers on that chart on average.

I like that worse. That would probably hurt their ability to spend money in 2012 and 2013, while they could simply continue with back-loaded contracts to whoever they sign in 2014-16. Money is worth more now. Effectively, by getting paid more in the less productive end of his contract and getting paid less in the more productive end of his contract, the Phillies have effectively borrowed money from Ryan Howard.

I'm not cherry-picking. Look through the list of players and their aging curves. It's not a perfect match, but the more extreme pull hitters decined more rapidly on average. Thome's WARP numbers would absolutely be satisfactory for Howard. Just 2005 looked bad. He was pretty sick in the other years. One bad year would not make the deal useless if he had a couple very good years or one spectacular year. There are a lot of paths that this could take. That's my point. I'm not saying it's a good deal. I'm painting the scenarios in which it works out well. There are a number of scenarios that aren't far-fetched.

That would definitely change things but I think the luxury tax treshold is $162MM in 2009, up from $155MM in 2008. That would be well beyond where the Phillies are headed.

Agreed. If anyone is playing the "I know your business better than you" card here, it's not me.

I have no inside information. I doubt Howard contributes too much to inflation. I think the net effect on the market is one less 1B, which creates scarcity, but one less team who needs a 1B, which negates that effect. I think it's a clue into what at least one team is thinking the market will look like.

I'm not quite sure what Black-Scholes would have added. It's been a decade since I actually did Black-Scholes, so forgive me if I'm misremembering, but isn't what I did equivalent to Black-Scholes with no time lag and a uniform distribution instead of a normal one? I doubt the added value would be worthwhile anyway, given the fact that these are only approximations anyway.

I discussed this more in the MORP series and it took some time so I won't go through the details. I largely based my estimates on a model developed by Sky Andrecheck at BaseballAnalysts.com. It is linked in the previous article. He does a terrific job of approximating career production by draft picks, and I simply rescaled it to look only at the first six years of contracts, arbitration money, and rescaling to fit WARP3 instead of WAR. There was a good deal of guessing that had to be done to determine which deals came at the cost of picks, but I was able to infer the answer pretty clearly from the outcomes in most cases. There were some judgment calls, but not major enough to change the $/WARP figure much.

Did you read the article or did you just show up to hate? As that section explains in detail, it's a sign that the Phillies are planning on spending money during that time period. That doesn't even mean it's smart for them at all. I like it when Starbucks buys comfortable chairs, as their customer, but that's separate from whether I find it wise for them to do so.

I'm not interested in the personal attack, so I'll ignore that part of the comment. As to the content of your post, I don't even know what to respond if you don't think that teams think winning and revenue are correlated, and don't consider whether they think they can be competitive when they set a budget. It's so far divorced from any anecdotal evidence about how owners of baseball teams spend money at all that it only reveals an inability to see budgets being part of any larger picture at all.

"Budgets" come about because firms first determine the cost function for different levels of revenue, and then select the level of costs that maximizes profits. Do they sit there with cost curves and their Econ 101 book? Of course not. But they behave in such a way that mimics this process. This is a very important fact about economic decision-making. The best analogy I've ever heard is to consider a billiards player. What they are actually doing in determining how to hit the ball is solving a ridiculously complicated series of differential equations. But they don't pull out Matlab and a Physics professor every team. They simply have arrived at a process of arriving at the same solution by learning over time how to succeed at billiards. Firms behave in a similar way. They don't take derivatives to solve every problem, but they seem to follow these principles. Owners typically know that the profits of a 81-win team and the profits of a 71-win team are similar, but the profits of a 91-win team are much higher than that of an 81-win team. Thus, when they feel they can only become a .500 team, they rarely spend money even in large markets like Washington. On the other hand, even smaller markets like Minnesota will spend money when they are on the sweet spot of the revenue curve. They do this by authorizing General Managers about what their budget should be. The result is that people think only in terms of the budgets handed down to the GMs and think this is all that's going on.

It's a mixture of all of the above. By there is a lot of revenue that comes in the form of merchandise, subseuqent years' ticket sales, and TV contracts as a result.

1) It does depend on the team that would have signed him. It could really be anywhere from about $18MM-$27MM depending on who signs him. My guess is that it would probably be a late first round pick, so $23MM seems about right. 2) If the Phillies don't pick up his 2017 option, it means that they didn't want to pay him $23MM; if they offer him arbitration, it means they will have to give him a raise on his $25MM. If the Phillies do pick up his 2017 option, they have essentially valued him at least as much as $13MM for 2017 ($23MM salary minus $10MM buyout), but offering him aribtration would require a raise on the $23MM figure, so it's also unlikely. 3) The cost of paying for lost wins in the form of dollars and picks is what I am counting here. I think we agree with each other on this point-- I determined the dollar cost of those picks and included that in the price.

Aging curves vary wildly from player to player. I think an analytic argument of aging curves should look at the selection bias of aging curves for re-signed players versus newly signed players when they are so dramatically different. The crux of my argument is whether the Phillies are right or wrong about his aging curve, and I give reason to believe that ignoring the fact that they watch him stretch and work out every day is dismissing very important qualitative information that is clearly correlated with quantitative trends.

As I wrote above, splits are fleeting. I'm not sure if you watched Howard last night but his "0 for 3" included a 400 foot fly ball to dead CF. The "late inning reliever" argument is overplayed, since teams only have so many LOOGYs. Greg Dobbs and Matt Stairs got a lot of PA vs. RHP in recent years because of this type of maneuvering. The "get one reliever and you're in the clear" argument is just not going to work. He still has a very high SLG vs. LHP, which is important in high leverage situations.

Pujols is going to get so much more than this that it's not even funny. If the economy improves and baseball keeps growing like it has been, I could absolutely see $40MM/year. I think the effect of one deal on the other is minimal, however. The opportunity cost of signing Pujols remains filling your team with a couple of 3-4 win players each at half his cost. Pujols' salary will depend on what teams are willing to pay him. Howard going off the market doesn't change what teams are willing to pay him.

Thanks. I don't know if hitting more HR to the opposite field is the reason. The best case I've made in the past is that hitters who are pull-happy rely on reflexes to hit HR, while hitters who are oppo-happy rely on strength. Reflexes decline more rapidly than strength, I think. It's just a trend I noticed. The larger factor, like you mention, is health. The best argument for Howard aging well has got to be that the Phillies know his health better than anyone, and they have one of the best medical staffs in the league (they won last year's Dick Martin Award from BP).

I agree that the market will depend on scarcity. The market always depends on scarcity. The reason that I unraveled a series of possible inflation alternatives is precisely because I agree with you that the market conditions are not set in stone and a number of possibilities should be considered.

My article used OPS to analyze his performance against LHP, which is already a quick and dirty measure. Your response is to use batting average? Ryan Howard is a low OBP, high SLG hitter against LHP historically. He's done better in his career against LHP than the average LHB. And most importantly, SPLITS STATS NEED TO BE REGRESSED TO THE MEAN IF YOU WANT TO PREDICT FUTURE PERFORMANCE. There's been massive amounts of research done on this topic. Also, if you want to get into "HR vs. LHP" argument, at least look at the rest of the league. Howard LED THE NATIONAL LEAGUE IN HR VS. LHP in both 2006 and 2007, and was third in 2008. It's a weak statistic, but at least compare him to other people. Nobody gets 650 PA against LHP.

The "how much talent is in the FA class" argument is somewhat misguided as well. What is lost in that analysis is that players on the FA market are removed from their old teams. Yes, that year's FA class will potentially also include Albert Pujols, Prince Fielder, and Adrian Gonzalez, but the Cardinals, Brewers, and Padres (or whoever Gonzalez is playing for at the time) will now all be without a first baseman in these scenarios. I'm not saying that the Phillies are forced to sign him. I'm saying that the Phillies have the altnerative of paying approximately the same price for wins as everyone else will at that time, so if they plan on being competitive, Howard may represent as good an option as any at the time. Under higher salary inflation, it could turn out to be an efficient way for the Phillies to get their wins. Anything near ten percent inflation with a modest rate of decline for Howard will make the cost of Howard's wins added lower than the alternatives on the market.

Market scarcity is already priced into MORP. That's the whole point of market values is that they reflect scarcity. MORP says the current cost of a win on the free agency market is $5MM including the opportunity cost of draft pick compensation. It could stagnate making this deal a disaster or it could inflate making it decent, but the concept of scarcity is not a new concept that needs to be introduce. It's the basis of market values. The Phillies don't use advanced stats to evaluate players as near as I can tell. They seem to be vaguely aware of them, but they are not using them in their analysis. Somehow, the groupthink on this has led to the conclusion that "if the Phillies don't use new school stats, they must focus on old school stats." That's ridiculous. The Phillies are a team heavily invested in scouting and health, which is the only reason that they've kept their heads above water while much of the league does better data analysis. I do not see any evidence this was based on RBI. I see evidence that the Phillies were focused on his workout routine and improved defense, and determined that he was unlikely to age poorly. If he maintains his value throughout this contract, it's a good deal. The question is about how much he declines, not about the stats which will decline. Also, there is a HUGE phallacy in the "wait for the market to develop" argument which I cannot believe has gotten this far. The market could go in EITHER direction. If the market gets better, Howard requires more money to re-sign; if the market gets worse, he requires less. This is about taking on a risk, which is generally smart to do in baseball given the shape of the marginal revenue curve. Put it this way-- if "waiting for the market to develop" was always a good idea for teams to do, it would always be a bad idea for players to do. Scott Boras would be in the poor house. His off-season strategy is tied up with hoping the market gets more expensive over the course of the off-season.

Yes, the opportunity cost of letting Howard go in 2011 is one of my main arguments as to why I think the value was higher than I'd have paid. If the next CBA eliminates this compensation before the 2012 draft, I think that the deal would look much better and probably wouldn't be over market value at all. Utley is around until 2013, and the Phillies will probably extend him after that. I think it's fairly reasonable that Howard will be able to assume the Phillies will put up some OBP in the 3-hole.

I think you are mixing up the definition of positive analysis. Positive = "what is", Normative = "what should be". I think you're suggesting I do a normative analysis and implying this is a positive analysis, but please tell me if I'm misinterpreting. This is a study about opportunity cost. In that sense, positive analysis is somewhat normative. What I am saying is "here is what teams are paying for wins" and therefore "good deals likely involve paying more than this." The reality is that without a team's actual books and projections of attendance and revenue under various scenarios, I cannot determine whether teams should be spending money on players in the way they do. All I can say is what the opportunity cost of a win is around the league, and therefore I can determine the point at which teams should choose to spend their money elsewhere.

PECOTA is a reference point. Never ever take PECOTA as the end of the story. If you look at the aging tables above, you will see that re-signed players do not age as PECOTA predicts, while players who sign contracts with new teams do exhibit this pattern. Additionally, I discussed the comparables that formed that projection, explaining why Howard's non-PECOTA characteristics have more in common with his more gracefully aging comparables. The fact that Howard hits better with runners on is something that is common among the majority of players who defenses shift against: http://www.thegoodphight.com/2009/1/29/741980/there-is-clutch-or-the-cas I wrote that article last year detailing this process. Howard hits more HR on the road, for one thing, and the stadium isn't going anywhere before 2017 anyway so he still gets to reap its benefits anyway. Joe Mauer was probably a great deal, though it really depends on his health. Still, it will probably look better than Howard's contract. Alex Rodriguez's will probably look worse if you check his PECOTA, but I'm guessing he will age more gracefully than that as well. Put it this way: Mauer was a bargain, but that wasn't the opportunity cost of signing Howard. Mauer was off the market. Other deals for this period of time are not yet signed. I don't think the contract is a great one, and I did say in the article it was a little high. I would have come in closer to $110MM or so. Still, just talking about PECOTA when I already looked at the PECOTA and decided where it was off is going backwards. Say why you think Howard will age like Cecil Fielder rather than Jim Thome. That would be a more compelling case.

This is a good idea. I definitely might do in the near future. I would need to think about methodology a little bit, but I think this would be a good thing to do. Thanks.

This is an approximation. The average draft pick that makes the majors debuts three years after being drafted. They play six years before reaching free agency, so the average WARP accrues about six years later. Changing it to 1/6 of the WARP for 4, 5, 6, 7, 8, and 9 years later probably won't change things much since 8% was a pretty noisy estimate in the first place. It's not precise because it's such a judgment call anyway. The goal was to get a more reasonable estimate of the opportunity cost of surrendering or forgoing a draft pick.

Yeah, you basically have it right there. It's true that pitchers do have control over ball-in-play outcomes, but the amount that they do control is relatively small and what they do control is correlated with strikeout, walk, and ground ball rates. Since SIERA is a regression analysis, it actually picks up on the portion of BABIP control that is correlated with those DIPS stats.

I think you might want to go back and check my series on Expected BABIP that I wrote a few weeks ago. It isn't really about explaining away performance, but there is a lot of luck involved and it can be a clue. In Jay Bruce's case, his infield pop up rate just wasn't high enough to justify such a low BABIP last year. His line drive rate was low, but those things tend to bounce back pretty quickly. Expected BABIP sees him as a .278 BABIP caliber hitter, which is below average but not extremely so. He loses points for having a swing that produces a lot of balls in the air, but his ability to hit for power negates a lot of that problem so that he's not so far below average. The kind of BABIP he had last year is the kind that you'd expect from a guy who has an uppercut but no power, not one with power. Bruce hits the ball too hard. In Hamels case, BABIP provided a clue that I analyzed a lot more thoroughly where it was just clear people weren't hitting the ball harder off of him. They were swinging and missing as much, fouling as much, hitting grounders/liners/flies/popups as much, and hitting the ball out of the infield just as much. BABIP provided the clue in that instance. Of course, it can't be the only tool but in the case of Hamels and even Bruce it's a pretty great way to get to the root of what changed.

You're more than willing that people are using statistical samples to making inferences, but than you seem to want to assume that all people who do so are suffering from confirmation bias. I'm just saying that people making inferences from statistical samples is a relevant aspect here. Unless you're unwilling to admit that some people are not inherently biased before they meet other people, then you shouldn't have a problem accepting this. People don't need to have statistics degrees to be statistically discriminating. I'm just talking about a general trend. It's hard to believe people put zero weight on any experiences they have ever. You're also totally missing the point of free entry. It's not that there are an infinite number of companies that need to consider entering. It's just that there needs to be no barriers to entry for any individual with a good idea. Very few industries are going to fail to meet this description. Baseball has a government sponsored monopoly and is therefore unique.

I agree that employer prejudice is a weak claim, which is why I focused on customer discrimination and statistical discrimination as more plausible in the article. Collusion can occur far more easily though when the set of firms (teams) is fixed in size without free entry. That certainly played a role in pre-Jackie Robinson baseball segregation. It's just that implicit collusion is harder to pull off from a baseline of integration. Certainly Americans can be harder to sign because of other obligations, but they are also subject to a draft which holds their available bidders down. That last argument might not be strong enough, especially because the discussion is focused on offers to existing African American players in general, rather than teenaged two-sport athletes.

You lost me with your first sentence. If you don't think rational behavior drives some of baseball economics after reading my columns, I just don't know what I can add in a comment that will change your mind. I think it's pretty obvious that the Indians had lower marginal values for CC, Lee, and VMart then their trading partners did and so there were gains from trade available where they had a higher relative benefit for future wins than present wins. Your example about Dye, Sheffield, and McCutchen isn't clear enough yet so you're going to need to elaborate. I don't think the existence of some players who risk saying no to contracts has anything to do with this issue. I'm also not getting the "quit whining" example. If Jermaine Dye were facing discrimination (which, again, I don't think is the case for him), that's millions of dollars that he doesn't get. That's enough for a generation of Dye to live off of, and so it's relevant to him. It might not seem like much when you have a lot of money until you realize that there is a generation of his family who will either be set for life or not based on a couple million dollars. That they like their jobs does not mean that they deserve to be ripped off, especially if it's inequitable.

Again, individual players have a variety of factors influencing their contracts. A player could have a questionable ticking time bomb injury that shows up during physicals that makes his value inexplicably low. It's tough to say definitively in one case. Looking at larger samples is just better.

I will probably try to come up with a good way to look at this empirically at some point, but it's a tricky concept so I'd need to solidify my approach to contracts in general once. It would be dangerous to try to find a one-to-one comparison like J.D. Drew because there are so many factors that go into an individual player's perceived value that you don't know when something else is going on. As far as the Branch Rickey aspect, I think that it was a risk that he was willing to take that others weren't. I think he probably thought it would be good for business, so in that sense, I'm somewhat in tune with the "market story" versus the "morality story." At the same time, it was obviously a risk that many owners would be on the fence about, and he went with the moral side and can be commended as such. Still, I think customer perception was a big part of the story, so I think Robinson's behavior was going to be crucial in any market story. I guess I'm saying that employer discrimination would be hard to maintain and it was only maintained as long as it was because of the lack of free entry, but that the customer discrimination persisted which is why even in 1947 only one owner took a risk on Robinson. I'm not sure there was enough information then for statistical discrimination to be playing a role. I think Robinson might have helped any statistical discrimination by being so calm and focused in the face of prejudice. If he created a bad stereotype, it would be harder for other players. Given the pressure he was under, it's tough to imagine many people being able to do so.

So why is it unreasonable to think that teams recognized Bradley's .999 OPS was an outlier? J.D. Drew has performed up to his contract overall, hasn't he? I don't think this is a good comparison. I also again fail to see where I'm misinterpreting Jackie Robinson breaking the color barrier. You mentioned lobbying above, but that's not related to what I was discussing. Mentioning various other aspects of the negotiations was not relevant to get my point across either. What I said is true, as far as I know. If I'm stating something inaccurate, please tell me.

Identifying the market inefficiency has always led to correcting it. If I have discussed a market inefficiency here that is researched further and identified, perhaps that will happen. I can't say. Hudson's value could be low for any number of reasons and it's not hard to see why people would take a shot at Milton Bradley anyway. Keep in mind though, as I mentioned earlier in the comments, that Bradley's last "valuation" on the marketplace was "=Carlos Silva". Not exactly getting away with his off-the-field issues right now.

I certainly might do some empirical research on this topic, but I did link to J.C. Bradbury's blog post which has links and summaries of a dozen different studies on the topic over the years. My point in the article, however, is that you need both theoretical and empirical research to talk about this issues because of other correlations. For instance, 57% of starting shortstops in the league right now were born in Latin American countries, compared to about 20% overall in the league. WARP and WAR treat positional adjustments differently, and if one of them is wrong, it will affect how we perceive Hispanic discrimination. The economic theory on this is not just a toy, and I was able to take it very far on this. Once you have theoretical modeling, you can do empirical testing better. I'm also not really sure how the existence of lobbying for Black players in the MLB discredits anything I wrote in the article. Obviously lobbying has an effect, but my point that Rickey took a chance and that Robinson had to endure customer discrimination and do his best to change it remains true either way.

What you describe certainly sounds unfair. I also am very doubtful that the wage gap-- which I think is something like 30% overall and 19% after you correct for education and experience-- can be explained by connections. That seems like way too large of an effect. I'm also pretty sure that Blacks earn less even controlling for parental income, but I haven't really tried to do that research since I was an undergraduate and I could be misremembering so please don't take my word on this.

I appreciate the comment, but please remember the distinction I made between discrimination and prejudice. If one makes inferences based on observed statistical tendencies, that is statistical discrimination but it is based on information even if it's a biased sample. The best example of statistical discrimination is the one I gave in the article about women's labor market exit rates. I do think that the market corrects issues when the affect is real. On the field production is very measurable, so I think the market corrects for these issues. I have always enjoyed the joke about the economist and the $100 bill on the street, but it's important to ask yourself if you have ever seen a $100 bill on the street. Probably not, I'm guessing. The dominance of the NL over the AL after the color barrier being broken fits right into what I'm saying in my article. The "magic of the marketplace" is due to free entry. As I explained in the article, free entry of new firms is not possible in baseball with finite teams, so it's possible to sustain a barrier like the color barrier. The bench filler issues that you bring up are a good example of where off-the-field value can push players over and under replacement level. The closer you are to replacement level, the more likely perceived off-the-field value can affect your ability to get a job. Please don't misunderstand my point about statistical discrimination. I am not justifying anything. I'm identifying the issue. If we can understand that the human mind categorizes and we can identify how it can categorize based on observations, we can work on solutions. Without solutions, it's just finger pointing. I don't quite understand why Hudson hasn't gotten more attention on the free agent market, but there are plenty of individual players who seem to be undervalued by the market (i.e. why did Cliff Lee twice get traded for lower value prospects than his contract and value would suggest?). That Hudson is Black can't be assumed to be the reason.

Milton Bradley is very uniquely talented, and even if his value is affected by off-the-field issues, he still would get quite a contract. Regardless, he was traded for no real value this past offseason because of off-the-field issues primarily, so the last record of his value is low.

Correct, Dukes is seemingly a terrible person and there is no issue with that affecting his actual and perceived value. I mentioned Myers as someone who has a domestic violence precedent, and who has at least showed up with a black eye one other time about which his story changed. Who knows why, but Myers is a questionable player but he's worth about one win and is compensated as such. Dukes is probably worth about one win and is not. There are plenty of players who get a mulligan, but I suspect that it might be correlated with race.

I don't think it's medical information in the way you're thinking. Players take physicals in all of these scenarios in general, so it's not really that. When I say medical issues, I mean just knowing how the player's body holds up day-to-day. Something like knowing how well a reliever's arm bounces back is something the player's old team just knows better.

I didn't use depth charts. I actually just used PECOTA spreadsheets. I'm sure that they overestimate IP on average but probably not for players who got multiyear deals since those players were less likely to be on the borderline between being in the minors and majors the first year. If PECOTA runs a little hot, that's another issue, but I don't think the CHONE system was running all that hot when it happened and Sean Smith didn't seem to see that as the issue in the article I linked.

I did look at age originally but it didn't seem to be the source of the bias, so I dropped that from the analysis. Players did age worse as they were older, but that just didn't seem to be what was going on at all.

VORP does account for league adjustments so that should not be an issue. I doubt that clubhouse chemistry or a change in time zones is the explanation specifically because the players who change teams are the ones who do worse and worse over the rest of the contract. The players who re-sign tend to underperform their VORP the first year, but then they don't really decline much at all. The players signing with new teams underperform their projections and then decline even further than that. That's why I think it has more to do with medical information as well as simple workout tendencies. For instance, I think the Phillies knew that Chase Utley was a workaholic and probably would have a longer peak than other second basemen so they signed him for longer.

In a one-sided test, 56% under-performances is significant at the 94.8% level (t-stat=1.63, with t=(93/165-.50)/sqrt(.50*.50/165). The argument is backed up by theoretical evidence though, which is very important here. There is a reason why it SHOULD be a biased sample. The additional evidence of the difference between aging of re-signed and newly-signed players is also evidence of this kind of bias. Also, this isn't 1/2 a win over the whole contract. It's just the first year. It gets worse than that for newly-signed players and better than that for re-signed players.

Correct. It is percent of total WARP for the whole deal in each year. It shows that players who re-signed maintained their value well but players signing with new teams did not.

I really don't think my cousin Mark knows who Javier Vazquez is. I'm bearish on Pettitte and somewhat bearish on Burnett. I think CC is probably a good bet too, but I think that Javier Vazquez has been the victim of bad luck and bad defense at the wrong times, and I think that the NL/AL league difference-- while huge-- is exaggerated to the point that people act like NL numbers are AAA numbers. Vazquez had a 2.68 SIERA last year to CC Sabathia's 3.70. I'm having a hard time thinking that the league difference is that big.

Also, your insult that I was referring to, which I'm repeating here is the implication that I have not thought about the point of this exercize which you called "academic." You again repeated that I was not applying logic. I think it should be abundantly clear that the whole strength of this exercize was in applying the logic. If you're not seeing it, that's a shame.

As I explained in the first article, the reason to regress BABIP separately is that K% and HR% are very reliable, and so you want to pin down the effects of the unreliable portion of batting average. The reason you won't do as well doing (outfield hits)/(outfield fly ball) as you will doing (outfield non-HR hits)/(outfield non-HR fly ball) is that hitters with difference home run skills are going to have vastly difference skills with respect to the former ratio but similar skills with respect to the latter ratio. Applying the same regression to all hitters on the first ratio will cause a huge problem. The point of discussing hitter BABIP was mentioned in the BABIP Roundtable we did a couple months ago, and it's well worth reviewing if you're still skeptical. The key is that you want to regress the statistics that represent more luck and less skill more than you want to regress the statistics that represent less luck and more skill. Home run rates have less luck than line drive BABIP. But infield hit rates have less luck than outfield fly ball BABIP. By breaking each skill down, you take the deviations from average that represent skill and regress those to the mean less than you regress the deviations from average that represent luck.

Last year, there was a .323/.298 BABIP difference at home vs. on the road for Colorado hitters, and a .311/.297 discrepancy for their hitters, so there is clearly some park effect, probably around .020 points. I'm not convinced that it's necessarily outfield fly ball BABIP that is getting the biggest effect, but probably some of it. Fowler's outfield fly ball BABIP probably is being regressed causing his overall BABIP to go down about .003-ish more than it should be regressed. The humidor may make balls sink quicker, causing more hits, or it could be that there is more space in the outfield than outfielders can cover. There are a lot of possible effects. I don't think defensive usage is going to move it much. How often do teams really rest an outfielder that would have failed to get to a ball that game and replace him with someone who would get to the same ball? A lot of stars need to line up for that to happen.

Definitely a good idea for an article. Thanks! As far as Kinsler himself goes, it's definitely his swing which is clearly a bit of an uppercut. He only had 32.1% ground balls last year (league average = 45%) and about 13.4% pop ups (league average = 7.8%). That seems pretty close to his old rates too. Numbers like that actually would imply a .265 BABIP or so I think, if not for his power and speed which bring him up to .279.

Tons, actually. Ryan Howard, for example, had a BABIP in the majors of over .350 through 2006, and even higher in the minor leagues. Then the shift came, and now he's projected for about .313, which is where he has been recently. Fortunately, that is already built into the projections for guys who have been shifted against. A young pull-happy lefty slugger might need his BABIP regressed as teams notice his pull-happy ways though, beyond even what E-BABIP would regress.

Hmm! Very interesting. I hadn't even thought about that fact. I would probably scale back what I said a little bit then. Texas' LD-BABIP last year was .708, and the league average was .729, so I'd bet you should cut short the "extra line drive BABIP" I gave Jones by .021, which would change his BABIP maybe .004. So maybe, he should have been around .255 last year instead of .260.

Definitely true. I think this would be more pronounced for hitters who are going to Boston. So, chalk up Adrian Beltre and Mike Cameron for a couple extra points for example. There also could be some over-regressing to the mean for some other hitters though. I think it would explain some small trends in BABIP certainly, but maybe we're talking about .005 or so for most guys.

The Colorado team BABIP on fly balls looks pretty normal, but the team BABIP on line drives seems pretty high. That could easily be one of Colin's classic press box heights or other scorer bias issues, though, and there could totally be more outfield fly balls that would fall in for hits anyway-- the size of the park certainly would suggest that. Clearly there are some park effects. However, Fowler is getting the same park effects in Coors in previous years as this year. If you just throw team=Rockies in as a dummy variable, you get about .009 of BABIP, but similarly for throwing it into old numbers. I would think maybe you've touched on something worth about .005 of BABIP or so for Fowler, but probably not enough to explain the .317 vs. .345 discrepancy. Very interesting thought, though. Thanks!

Now I'm not getting the link to work at all. Try this: http://spreadsheets.google.com/pub?key=tMC8mqXiVL6gP9bnd74jrEw&single=true&gid=0&output=html

I meant bracket, not quotation. So there's a bracket, a slash, and the letter A at the end of the hyperlink. Delete those, and that should work.

There's a quotation, a slash, and an A at the end. Somehow even typing that into comments cuts it off. Just delete those three characters. I've emailed about this, so hopefully it can be corrected. Thanks and sorry!

Looks like there is an extra three characters "

From yesterday's article: "They also were closer to actual BABIP than PECOTA was 57 percent of the time, and closer to actual BABIP than CHONE was 60 percent of the time. That might not seem like much, but those fractions are significant at the 95-percent and 99-percent level, respectively. In other words, there is less than a five-percent chance that there would have been that large of a difference between my BABIP model and PECOTA if they were equally as good, and less than a one-percent chance that the model would have beaten CHONE as badly as it did just by chance." Implicit in this is that if you add in the same number of HR and SO that PECOTA predicts, but change the number of hits and outs on balls in play to E-BABIP, you will have a superior prediction of AVG. If you're going to make flat out accusations and hurl insults, you should at least do a brief scan of the articles to make sure I didn't address this. I also talked about it in last year's "BABIP Superstars" article and last year's "You Can Beat PECOTA Without a Computer Model" articles.

I think you mean Baseball-Reference, don't you? I can't find it on Fangraphs. I can't get any normal output that I could automatically merge, and I haven't been able to get reliable massive datasets in a form that wouldn't involve typing every number in for each individual player. In truth, I stink at programming, so I'm reliant on excellent help from Eric Seidman and others for these things. I'm not sure if people here are comfortable replicating the B-R L/C/R numbers reliably without muddy data.

How can you honestly read these two articles and actually type your Step 1? Do you realize that balls in play being hits more often will necessarily imply higher batting averages? Of course players who have high BABIPs will have higher AVG and higher OBP. Studying BABIP as a subset of performance clearly has value, as I've explained in both of these articles, because it is more complicated to project. The point of projection is to say that Michael Young and Joe Mauer will not only have "highish batting averages" but to pinpoint how high. To the extent that you can increase the accuracy of how high you think it will be, projecting BABIP has value. Do you believe projection is a useful concept at all?

All fly balls are included, but the use of HR/FB is mostly a proxy variable for power. It's not implicitly subtracted at all. The more frequently hitters hit 400 foot fly balls, the more often they hit 350 foot fly balls so the more ground outfielders have to cover, and the fewer balls they catch. There's also another reason, though. Say you have 100 balls in play and that there are 40 ground balls, 20 line drives, 30 outfield fly balls, and 10 pop ups. If none of those are home runs, then line drives only make up 20% of your balls in play. However, if you're Prince Fielder and 10 of every 30 outfield fly balls leave the yard, then 20 of every 90 balls in play are line drives-- 22% instead of 20%, which would drive BABIP up by about .010. So there's two effects there.

My R^2 for the 3 years regression was 0.31, not 0.20. The reason that you are getting weird P-stats for GB% FB% and LD% is because of multicolinearity. They should all add up to 100% because Fangraphs includes Pop-ups and Outfield Flyballs in "Fly Balls", so one should be removed from the regression. One of them would actually have automatically been removed if not for I'm guessing Fangraphs' numbers are rounded and so they might theoretically have added up 99.9% or 100.1% for some people. Pick one, take it out, and leave the other two in. Rerun the regression. Your R^2 is probably a little higher because you used Fangraphs' default PA which is higher than 300. That makes your R^2 more exact for your dataset. Running mine on the same data set by restricting it to 500 PA gives me an R^2 of .38, for example You're actually running almost the exact same information as I am, minus GB-BABIP, OFFB-BABIP, and CONTACT%, so you're going to get most of the way there anyway. It sounds like you did. Standard Deviations of 18.5 points sound almost perfect, as it should, and naturally that means 65% should fall within 18.5. Running the regression with an intercept of 0 was an unnecessary sidebar. Setting the intercept equal to 0 means requiring things to all be 0, what is the variable. So for some who theoretically has 0% GB, 0% FB, and 0% LD and 0% IFH, 0% IFFB, 0% HR shouldn't even have a BABIP. And doing so while dropping one of the three collinear variables (GB, FB, and LD) would be arbitrarily assuming BABIP for some with 100% of that batted ball type is 0.

I'm not taking them out of the analysis. I regress on GB/BIP, PU/BIP, LD/BIP. That implicitly assumes OFFB/BIP is included. You can't included four things that add up to 100% in a regression because of multicollinearity. Instead, you include the three other things. The reason ground balls and line drives are positive coefficients is because the more of them you get, holding popups constant, the less flyballs are hit. And ground balls and line drives have higher BABIP than outfield fly balls. Similarly, popups are negative because holding line drives and ground balls, constant, more popups and fewer outfield fly balls means a lower BABIP.

Thanks for the compliment :-) It's a good question. I would guess it's more useful at higher levels than lowel levels, and you would probably need to come out with some sort of Minor League Equivalencies for each metric to convert it. Also, since ballparks are so different across the minors, you couldn't get away without doing park effects as I've attempted to do. Once you figured out the MLEs and PFs, you could probably do well to convert infield hit rate in the minors to the majors considering the 3B are better, convert outfield fly ball BABIP to reflect better fielding OF. The real question is how much line drive rate would carry over. I would guess that although major league pitchers don't have much difference in their LD% abilities, there is probably a lot more difference between the LD% of MLB pitchers and pitchers in high-A ball. If you converted all these things, you really might be able to do something like that assuming you were careful. At a more qualitative level, you could probably do a lot. What is the groundball/flyball ratio of this guy in AAA-- does it indicate a downward plane to his swing? Does pop-up rate shine any clues? How often is he reaching on infield hits? How much power does he have? What's his contact rate? How well does he get ground balls to the outfield and fly balls to fall in the outfield compared to other hitters with similar skills? Is this indicative of a good ability to spread the ball around? Once you know all this, you can probably pin down where a hitter is. A speedy ground ball hitter who spreads the ball around, makes good contact, and rarely pops up might be a .330 BABIP kind of guy. A power hitter who rarely hits ground balls but spreads the ball around well might be a .285 guy. A power hitter who pulls everything but hits everything very hard and rarely pops up might be a .315 guy. Things like that are very informative.

Thanks for the question. There are several reasons for this. The primary reason is that ground ball hitters are good at BABIP, but not an EqA or any other measure of overall offense. Take Juan Pierre. He's a below average hitter with above average BABIP. The reason is largely related to ground ball rates. Another large factor is pitchers do not control their rate of home runs per outfield fly ball. So pitchers who allow more fly balls naturally allow more home runs. However, not all hitters that have high fly ball rates are going to have high home run rates, because there are huge difference in the rate of home runs per fly ball for hitters. My article last week on SIERA and BABIP actually showed that pitchers who allow more fly balls have lower BABIPs, which is partly why the "cost" of a fly ball isn't as high for a metric like SIERA as it is for other metrics that are based on run estimation of individual outcomes. Since pitchers who allow more fly balls also allow more home runs, a fly ball is a bad thing for run prevention in both metrics, but it's not as bad because SIERA implicitly assumes that their BABIPs also go down a little. I guess the real difference is home runs. BABIP is part of hitter skill, but certainly not all.

It definitely is. THT's xBABIP uses this statistic which they call "spray". Foruntately, using historical BABIP rates seems to be picking up this effect for me, but if I could get my hands on that data, I would like to see if high historical BABIPs on grounders and fly balls are more likely to be maintained by hitters who spread the ball to all fields.

I'm not sure if Marc still uses that model, and what for, but it's been debunked as a predictor of next year's BABIP. The correlation between BABIP in year-1 and BABIP in year-2 is higher than the correlation between (LD% + .120) in year-1 and BABIP in year-2. The reason is predicts same year BABIP pretty well is that BABIP on line drives is about .730 and on everything else it's about .190. The difference between the best and worst at BABIP skill is only about .680-.780 in skill level and the difference between the best and worst at non-LD BABIP is probably only .140-.280. So if you weight those two groups, you can always get close IF YOU KNOW THE % OF LINE DRIVES. The real problem is that line drive has only a .37 year-to-year correlation, making it relatively less accurate than knowing a player's speed, power, contact, and GB, OFFB, and PU rates.

Okay, I'll pick a few that might be harder to guess: Casey Kotchman, Nick Punto, Martin Prado, Lyle Overbay. Take a look at those and see if you can guess whether I think PECOTA is high or low on them. They'll be in the article tomorrow along with 6-7 others IIRC.

Thanks! That will be at the end of tomorrow's article. Tomorrow's article is going to talk about E-BABIP vs. PECOTA-BABIP. Then I'll link to the E-BABIP spreadsheet at the end.

Thanks. The thing is that line drive rate DOES have a high correlation with BABIP that year, but line drive rate is not very persistent. So certainly a guy who has 19% line drives (league average) might have a BABIP of .300 (league average, holding everything else average), while if the same guy had 24% line drives, his BABIP would be .325. However, a guy with 24% line drives one year is probably going to have a 21% line drive rate the next year, which might correspond with a BABIP of .310. So 5% more line drives in 2009 indicates only .010 points more of BABIP in 2010.

Check out the SIERA and BABIP article I wrote last Wednesday for a more detailed answer, but pretty much I think that pitcher BABIP is almost entirely explained by K, BB, and GB skills.

I think 4-5 wins with 10.5% BB is exactly where he's been lately. I guess you're probably guessing his HR-rate falls. That's certainly possible but Thome seemed to maintain his pretty well. I think there are different muscles and reflexes involved in hitting oppo-HR, and those just tend to be muscles and reflexes that age better. We really know so little about aging. In this article (http://baseballprospectus.com/article.php?articleid=9055), I talked about how sluggers who hit opposite field HR seemed to age better, at least based on my puny sample size of Howard's 2009 PECOTA Card comparables.

xBABIP is a post-dictor/estimator, rather than a predictor of BABIP, which is a useful concept certainly (SIERA is a post-dictor/estimator of ERA), but serves a different purpose. More specifically, it uses same year variables to predict that year's BABIP. E-BABIP predicts future BABIP based on batted ball breakdown. A lot of the original xBABIP model has been changed due to suggestions they got at the time. For instance, they realized that using home run rate to model power was better than pitches/XBH, mostly because it was not putting doubles and triples directly in the set of independent variables when it's supposed to be the dependent variable. The speed score is supposed to represent ability to get hits on infield ground balls or draw the infield in. However, to post-dict BABIP, you can't just include infield hits because obviously that's including some singles in the independent variables already. (That would be like including RBI doubles in an ERA estimator-- of course you're ERA is higher, but really that's not what you're measuring.) However, to predict BABIP, it's okay to use historical infield hit rate, which is what I use (+ reaching on errors). Then you're simply asking how well the past predicts the future. I did use handedness in my regressions originally, but it did not do more than add or subract .001 from overall BABIP and wasn't significant. The reason is that any of the ways that handedness would affect BABIP would have already affected GB-BABIP and OF-FB-BABIP in previous years. xBABIP was trimmed down a lot and adjusted later. The things like discipline had been introduced, and things like home run rate and pop-up rate were things I was already using in older models, and the lefty*(gb/fb) variable is really just modeling infield hit rate too. The real add-on, and data that I'd love to get my hands on and incorporate meaningfully, is that they introduced "spray" which basically measures how well the hitter sprays the ball across the field. That would be useful to improve this too. I think that's the main contribution of xBABIP is proving that a wider distribution of batted balls across the field plays a major role in BABIP.