CSS Button No Image Css3Menu.com

Baseball Prospectus home
  
  
Click here to log in Click here for forgotten password Click here to subscribe

<< Previous Article
Premium Article Future Shock: Further ... (04/21)
<< Previous Column
Premium Article On the Beat: Cooper's ... (04/19)
Next Column >>
Premium Article On the Beat: Setbacks ... (04/26)
Next Article >>
Prospectus Hit and Run... (04/22)

April 22, 2009

On the Beat

Reeling in the Marlins

by John Perrotto

the archives are now free.

All Baseball Prospectus Premium and Fantasy articles more than a year old are now free as a thank you to the entire Internet for making our work possible.

Not a subscriber? Get exclusive content like this delivered hot to your inbox every weekday. Click here for more information on Baseball Prospectus subscriptions or use the buttons to the right to subscribe and get instant access to the best baseball content on the web.

Subscribe for $4.95 per month
Recurring subscription - cancel anytime.


a 33% savings over the monthly price!

Purchase a $39.95 gift subscription
a 33% savings over the monthly price!

Already a subscriber? Click here and use the blue login bar to log in.

The Marlins were considered an afterthought in the National League East when the season began. No one doubted that the Marlins would do what they normally do, scrapping and overachieving despite tight-fisted owner Jeffrey Loria saddling them with the lowest payroll in the major leagues at $36 million, and it seemed somewhat far-fetched to think that the Marlins could contend in a division that includes the defending World Series champion Philadelphia Phillies, the New York Mets with their revamped bullpen, and the Atlanta Braves with their new-look starting rotation.

While just over two weeks of action represents a meaninglessly small sample from a single season, it's large enough for the Marlins, who believe that they can win their first division title in franchise history. (Such is the oddity of modern baseball: the Marlins have won two World Series titles since making their debut as an expansion team in 1993, but both came after they had qualified for the postseason as the Wild Card.) The Marlins don't think that there is nothing unusual about their 11-3 record, the best in the major leagues. The fast start has given them reason to believe they coud be playing meaningful games deep into September.

"We respect the Phillies and the Mets and the Braves," Marlins outfielder Jeremy Hermida said. "They deserve to be mentioned as the favorites in this division, because they've done it before and we haven't. That being said, we believe we can play with anyone in our division. We feel like we can contend. You don't expect to start off the way we did, I don't think anybody thinks you're going to go 11-1 to start a season [before losing to the Pirates on Monday and Tuesday]. At the same time, it's not a total surprise. We believe we have a good team, and that we're going to show over the course of the season and not just for a couple of weeks. We're not some kind of fluke."

The Marlins are fifth in the majors in runs scored per game with an average of 5.9, and sixth in runs allowed with a 4.1 average. "What I like about the way we've played is that everybody has shared the duty," Hermida said. "It hasn't been one guy carrying us, or the just the offense, or the pitching staff. It's everybody doing their part, and that's what we have to do. We don't have many superstars, so we all need to pitch in."

Even if the Marlins did have a batch of superstars, who would notice? No one goes to their games in South Florida, which is understandable since they play in a football stadium and rarely make a national television appearance. While shortstop Hanley Ramirez finished second in the major leagues in VORP last season behind Cardinals first baseman Albert Pujols, the highest-ranking Marlins player on that list so far in 2009 is first baseman Jorge Cantu, whose 4.9 is 61st.

While the Marlins feel that their offense will generate enough runs, they also realize that their chances of remaining atop the NL East rests with their young starting rotation, with right-hander Ricky Nolasco being the elder statesman at 26 in a unit stocked with such talents as Josh Johnson, Chris Volstad, and Anibal Sanchez. Johnson had Tommy John surgery on his elbow, and Sanchez underwent shoulder surgery during the 2007 season and did not return until last July.

Marlins manager Fredi Gonzalez believes that this year's fast start actually began with the return of both pitchers. "Having those guys come back last season and prove they were healthy gave both of them a big lift coming into this season," Gonzalez said "They are both on top of their games now. When you add them to the pitchers we already had, it makes for a good rotation. We're an organization that prides ourselves on developing or acquiring good young pitching, and ultimately how well we do this season is going to be determined in large part by how our starters do. I have a lot of faith in our rotation. They may be young, but they're good pitchers, and I feel we have a good chance to win every night because of them."

While some sabermetricians might downplay the concept of karma helping a team to win, the Marlins can't help but wonder if perhaps this is their year just based on what happened last weekend against the Nationals in Washington; they became the first team in major league history to sweep a three-game series in which they trailed in the final at-bat of each game. "I think you put something like that away, and count on it when times get tough," Gonzalez said. "As well as we've played early this season, there is going to be that stretch where we lose five or six games in a row. It's inevitable. It happens to even the best of teams. When we hit that streak, though, we can think back to the weekend in Washington, and that will give everybody a little boost of confidence. We'll know that, regardless of how bad things might become, they can always change for the better very quickly."

---

Twenty-one home runs have been hit in the first five games played at the new Yankee Stadium. The House That George Built has looked like a launching pad in the very early going, as no other stadium in major league history has ever had more homers hit in the first five games (Cincinnati's Great American Ball Park yielded 20 in 2003). "It's definitely playing short," Yankees general manager Brian Cashman told John Harper of the New York Daily News. "We're trying to get our arms around the reason for it."

Accuweather.com meteorologists told the Daily News that home runs are most likely to be hit when the wind is blowing from the west at 10 mph or more, and those conditions usually occur in the Bronx in the spring and fall. It stands to reason that if that projection holds, home runs should begin to tail off around one-third of the way through the season.

If that's the case, then Yankees manager Joe Girardi believes that the new Yankee Stadium will play much like Wrigley Field in its wind-driven seasonality, the park he called home from 1989-92 and 2000-02 as a catcher for the Cubs. "You managed there differently in April and May than you did in July and August," Girardi said. "That's why I think we should wait and see before we determine exactly how this stadium is going to play."

Indians reliever Jensen Lewis doesn't need any convincing after serving up the game-winning home run to Yankees catcher Jorge Posada last Sunday. "I thought it was a routine pop fly," said Lewis. "In any other ballpark in the country it's an out, but here it gets into that jet stream and it's out. I don't think the numbers lie. You see all these balls flying out of here."

---

Rangers manager Ron Washington's days have seemed to be numbered ever since Nolan Ryan became club president prior to last season, but Ryan made it clear this week that Washington's job is not in danger, and he has not considered switching skippers. "That's only through the media," Ryan told Anthony Andro of the Fort Worth Star-Telegram. "That's something you'll have fabricated. There have been no discussions or anything about that among the front-office staff. When I hear people come up and ask me about that, it has to come from somewhere."

The Rangers are 6-7 this year after going a combined 154-170 in Washington's first two seasons. "I think Ron's fine," said general manager Jon Daniels. "I don't see anything that concerns me. I think he's done a good job. We've had some well-pitched games where we didn't hit and score any runs. Then we had some games where we had a breakdown once we went to the bullpen, or the starting pitching didn't give us what we wanted as far as innings. As a whole, I'm pretty happy with where we are. We're in a good state of mind."

---

The Cubs/Cardinals rivalry may not generate the same level of attention from the national media that the Yankees and Red Sox do, but it is always contentious, and the heat was turned up a notch when the teams played last weekend. Cubs first baseman Derrek Lee questioned whether Cardinals pitcher Adam Wainwright had received the benefit of the doubt from umpire Larry Vanover on a called third strike to the Cubs' Milton Bradley, who became so enraged during his argument that he wound up receiving a two-game suspension from Major League Baseball.

Lee's suggestion that the Cardinals were getting help from the umpires did not sit well with St. Louis manager Tony La Russa, who said, "I don't how the Cubs get away with the comments they make about umpires."

Cubs manager Lou Piniella grew up with La Russa in Tampa, and he laughed off his old pal's remarks. "Tony's a good friend and a heck of a manager, but he is not the commissioner," Piniella told Dave Van Dyck of the Chicago Tribune. "Not yet, anyway."

---

Major League Rumors and Rumblings: There doesn't appear to be anything to the rumors that the White Sox might bring Aaron Rowand back to play center field in a trade with the Giants, but there is growing sentiment within the organization to move shortstop Alexei Ramirez to center and call up shortstop Gordon Beckham from Double-A Birmingham. ... The Athletics continue to have interest in free-agent pitcher Mark Mulder, and the Dodgers and Nationals are also considering him. ... The Diamondbacks have offered a front-office job to former outfielder Luis Gonzalez, who spent last season with the Marlins, but he has put them on hold because he's still hoping to sign with someone as a free agent. ... Pedro Martinez continues to sit on the free-agent market and wait for a call from a pitching-needy team.

---

Three series to watch this weekend, with probable pitching matchups:

John Perrotto is an author of Baseball Prospectus. 
Click here to see John's other articles. You can contact John by clicking here

Related Content:  The Who,  Marlins,  The Call-up,  Marlins Park

36 comments have been left for this article. (Click to hide comments)

BP Comment Quick Links

Fresh Hops

Are we sure that the winds in the new Yankee Ballpark aren't a monsoon called Chen-Ming Wang?

Also, I disagree that "While just over two weeks of action represents a meaninglessly small sample from a single season." Real wins are never meaningless, even if they don't change our estimation of a team's true talent. Suppose you still think that the Marlins are a .444 team. Well, they have 148 games left to play and that means 65.7 wins within those. With 11 wins under their belt, you should think that they will win 77 games, or five more than you thought before. For a team everyone expected to be as bad as the Marlins, this early run isn't enough to make them contenders. But for a team like, say, the Mariners, winning the AL West is now a serious possibility simply because a .480 true talent team that's now probably going to finish above .500.

Apr 22, 2009 09:22 AM
rating: 1
 
HeavyHitter
Other readers have rated this comment below the viewing threshold. Click here to view anyway.

Hey philosofool, ever hear of "regression to the mean?"

Apr 22, 2009 09:56 AM
rating: -11
 
Dan

philosofool is exactly right. These games that are in the books now may not change the true level of a team, but wins are wins. They don't take them away from you.

If the Marlins are really a .444 team, they're not expected to be WORSE than .444 just to get their overall record back down to .444. They'd be expected to regress to their level and play .444 ball the rest of the way. When you tack an 11-3 on the front of that, they'd wind up at about 77-85, as he said, and that's more like .475.

(Now, is their "true level" really .444? That's a tougher question.)

Apr 22, 2009 10:19 AM
rating: 1
 
Patrick

Actually, if the Marlins are a .444 team, they would play worse than .444 to end up at that number. Think of it like a coin flip. You could flip 11 heads and 3 tails in the first 14 flips, but by the time you get to flip 162, you're likely to be closer to 81-81 than 85-77.

Apr 22, 2009 10:46 AM
rating: -3
 
eighteen

No, no, no, no, no, no.

We're talking independent events here. Prior outcomes don't determine future outcomes, nor change probabilities. The remaining 148 flips are likely to be 74-74, or very close. The 11-3 start doesn't change that. By the time you get to flip 162, you'll be closer to 85-77 than 81-81.

Apr 22, 2009 11:03 AM
rating: 4
 
Patrick

Right, but I'm looking at all 162 flips, not just 148 with 11-3 tacked on at the beginning. The 11-3 start does makes it more likely that your coin flips will end up at 85-77, but as you increase the sample size, the ratio of heads to tails will even out, necessitating more tails than heads as you get closer to 162. True, it might not happen, but the ratio will approach 1:1. It doesn't matter how you get there.

11 wins in 14 games (out of 162) isn't THAT unusual and says very little about the Marlins being better or worse than a .444 team.

Can we get a statistician to weigh in on this? It's been nearly ten years since I had to think about this kind of problem.

Apr 22, 2009 13:05 PM
rating: -1
 
Adam Hobson

But you aren't looking at 162 future flips. You are looking at 14 flips that already happened with an 11-3 outcome. Now there are 148 future flips to come, which are more or less independent of the previous 14 future flips.

But that leads to an interesting question, just how independent are individual baseball games? I'd definitely think that hey were far more independent than dependent, but not completely.

Apr 22, 2009 13:51 PM
rating: 1
 
Patrick

My point is that if the Marlins truly are a .444 team, as the sample size increases, they will approach .444, which would require them to play worse than that. If they play .444 ball over the rest of the season, wouldn't that make them a .475 team?

Apr 22, 2009 15:10 PM
rating: -1
 
Adam Hobson

I think that's kinda the point. .444 is just the projection. Now obviously most teams are not going to hit their projection exactly. But that's kinda the point. If a team is projected as a .444 team and starts the season at 11-3 it's already beating that projection, and is more likely than not to to beat the final projection as well, since it's already a few games up over the projection. Of course that doesn't mean the team will definitely beat that projection, just that it will be more likely to do so now.

Apr 22, 2009 21:58 PM
rating: 0
 
Ben Solow

Patrick, I'm not a statistician, but I am getting my ph.d. in a field that uses a lot of statistics (I'm studying to become an applied econometrician). What you're talking about are asymptotic properties of estimators -- how the estimator behaves as we make the sample size very large -- and the law of large numbers specifically. The law of large numbers says that as you increase the sample size without bound (mathematically speaking, take the limit as sample size -> population), the moments of the sample converge to the moments of the distribution. Moments in statistics are defined as mean, variance, skewness, kurtosis, etc., and you are referring to the mean (winning percentage). Unfortunately, a 162 observation sample is not a very large sample, and maybe not enough for the law of large numbers to hold extremely accurately.

A better thought experiment is this one: assume that if you played a nearly infinite number of games, the Marlins would win 4/9 of them. Right now they are 11-3, and we play until we've played 1-million games. Over the next 999,986 games they win 4/9 of them. That means the Marlins would win 444,438 and lose 555,548, bringing their 1-million game total to 444,449-555,551, which is nearly indistinguishable from their expected 1-million game total of 444,444-555,556.

In other words, while we should expect a team to approach their true talent level asymptotically, there's no reason that it should occur over a 162 game sample.

Apr 23, 2009 16:14 PM
rating: 1
 
Adam Hobson

"My point is that if the Marlins truly are a .444 team, as the sample size increases, they will approach .444, which would require them to play worse than that."

I just reread that again, and I see where the confusion occurs. Regressing to a .444 team doesn't mean that they well end as a .444 team. Right now, at a 11-3, the Marlins are a .786 team. If they play at .444 the rest of the way, they will end up as a .475 team. By going from a .786 team to a .475 team, they still regressed to .444, they just didn't reach it all the way. And like Ben said, if they play another 162 games, or another million games, that eventual winning percentage will get closer and closer to .444. That's what regression means, not that they will be expected to reach that number, but that they will get closer to it.

And of course since we are just talking about probabilities of future events, they could actually reach .444 or even lower, there's just less likely of a chance that they will do so than end as a .475 team.

Apr 23, 2009 19:43 PM
rating: 0
 
jlefty

Prior outcomes don't determine future outcomes, nor change probabilities.

Correct. But was BP projecting that in any given game the Marlins have a .444 chance of winning? OR that at the end of 162 games they will finish with a .444 win percentage? Obviously a team's chance of winning any given game isn't static, therefore, I'd say the latter is correct. It doesn't matter what the Marlins do in the first 2 weeks (especially playing against the Marlins), BP sees them as a .444 team at the end of the season.

Apr 22, 2009 14:44 PM
rating: 0
 
Adam Hobson

Ok, that's partially correct. The Marlins don't have a .444 chance of winning an individual game, because obviously the particular opponent and starters will change that probability around a bit.

However, BP saw the Marlins as a .444 team at the end of the season before the season started. That means (unless they change that prediction due to new information) that the Marlins would still be predicted to be a .444 for the remaining games on the schedule as well, with a small caveat.

That caveat would be if they were more like to win the first 14 games, which is likely considering that many were against the nationals. I think we can assume that the Marlins would not have been predicted to go 11-3 for the first 14 games, even if those first 14 games had a higher chance of victory (thanks to strength of schedule) than the last 148 games.

So let's say that they were expected to go .500 for the first 14 games, which means they would be expected to go .439 for the remaining 148. However, in reality they actually went 11-3 over the first 14, which doesn't change that they would still be expected to go .439 the rest of the way. That leads to 76 wins (instead of the projected 72) for a .469 winning % (instead of the projected .444)

So those wins that actually already happened still all adds up to a greater winning percentage at the end of the season than predicted.

Apr 22, 2009 14:59 PM
rating: 0
 
jlefty

But my point is that BP sees them as a .444 team at the end of 162. So you can't split the season at any arbitrary point (e.g. what has happened so far in real life) and say "they'll go .444 the rest of the way."

It was a final projection, so (if you accept it as correct) you have to expect the Marlins to play worse than .444 from here on out in order to hit the projection.

If you think the 11-3 start indicates that the projection WASN'T correct to begin with, then why believe that it will be correct over the course of the rest of the season??

Your methodology almost makes it IMPOSSIBLE to ever correctly predict a season. Suppose the Marlins were 1-0 (not 11-3). And you use your methodology to determine they will go .444 the rest of the 161 games. That's 71.55 wins. For a total of 73. So because they won their first game rather than lost it they won't be a .444 team at the end of the season? Conversely, a team projected to end with a winning record couldn't lose their first game, otherwise your method would again show the projection to be wrong.

Apr 22, 2009 15:31 PM
rating: 0
 
R.A.Wagman

It is impossible to correctly predict a season. Absolutely and utterly impossible. It is possible to make a reasonable and educated estimate based on trends - this is what BP does.

Apr 22, 2009 18:13 PM
rating: 1
 
jlefty

...seriously?

I'm pretty sure if the Marlins finish with 72 wins, BP will have correctly "predicted" their win percentage. I beg you to understand that I'm not using the word "predict" too literally. You knew that, right? Ok.

Now. BP expects the Marlins to have a .444 win percentage. The argument here is how their record so far affects that projection. Do they finish the season that much over .444? or do they play worse than .444 for the rest of the year and hit the projection? (assuming the projection was correct to begin with).

Since the projection would have been proven wrong after the Marlins won game 1, I submit that BP's projection remains unaffected by the hot start and that they will regress towards their projection.

Apr 22, 2009 21:57 PM
rating: 0
 
Adam Hobson

So first, here's a question to you: how does winning 11 games in the first 14 affect the rest of the season so that the rest of the season their winning percentage would be lower? How does a win in April turn a win into a loss in July?

The .444 winning percentage was just a projection. That's basically saying we think the talent level of that team will lead it to win .444% of the time. Winning 11 games of 14 dosn't change that talent level for the future. Future games are independent of past games. So even the Marlins are 11-3 now, their talent level is still that of a .444 team, so that is what they should do going forward.

So yeah, the projected .444 team may end up with an actual record of .475, or maybe .500 or maybe .430, but that's because the .444 was just a projection, not a vision into the future. And yes, the Marlins could go .439 the rest of the way to end up with 72 wins and .444 winning percentage, but if their talent is truly at the .444 level, their is a lower chance of that happening.

The key here is to properly understand the difference between dependent and independent events and probability. What you are describing is similar to the Monty Hall Problem/Paradox, but the difference there is that the events are not independent. In baseball, future game outcomes are more or less independent of past game outcomes.

Apr 22, 2009 22:12 PM
rating: 1
 
jlefty

I understand the difference between dependent and independent events just fine, and I know a thing or two about probability, trust me. The key here is to realize that these are BASEBALL GAMES, not coin flips. Each and every flip of a quarter has the same exact odds, and therefor if in the first 14 flips there were 11 heads, we could expect there to be more than 50 heads at the end of 100 flips. But that's dealing with a static, known probability.

The odds of winning a given baseball game in a 162 are certainly not static, and obviously not known. So, you can either accept BP's projection of .444 or reject it. But, if you accept it at first, under your methodology, you must then view it as incorrect after they start the season 1-0. Does this seem reasonable to you? Do you think that to project a team to go .444 is equivalent to saying "they will definitely lose their first game?" (or for that matter "they will never have a stretch of 11-3 over 14?") And again, if the first game of the season can make you think the projection was wrong to begin with, then why think that it will be right over the next 161?

Or even on the last day of the season...are you going to abandon the projection if the Marlins are 71-90 with 1 left to play? My basic point is: if you think the projection is accurate, then you think it's accurate. The projection is over 162 games. you can't say well it was wrong over the first x, but it will be right over the next 162-x!

The reason for the low projection to begin with was the Marlins facing the toughest schedule out of any team in the majors. So far, they've had one of the easiest. Is it so hard to imagine that they'd play better than their final projection over an easy part of the schedule and worse than it over the harder part?

Apr 23, 2009 06:31 AM
rating: -1
 
Adam Hobson

I think you are putting too much stock in the actual projection. The projection is a guess, a conjecture, it is not a vision into the future. Most projections are wrong, whether than be by a game or more. So I have no problem that a projection projects to be wrong at 14 games into the season, or even just one game.

Let me explain this from a more extreme real life example. Consider Chien Ming Wang. He's thrown perhaps the worst six innings in baseball history in his last three starts. In this six innings Wang has given up 23 earned runs to a tune of a 34.50 ERA. Now Pecota predicted Wang to sport a 4.28 ERA at seasons end. Wang's next start is being skipped over due to an off day, and he's been sent to Tampa to work on his mechanics with the possibility of a DL stint. Let's say that he comes back from that with either his unknown injury fixed, or they found a flaw in his mechanics and corrected it so his sinker sinks again. So now he comes back as the same pitcher Pecota predicted to finish the season with a 4.28 ERA. Except now thanks to the six innings and 23 earned runs that already happened, he's just not going to finish the season with that 4.28 ERA. Just because he threw the worst six innings in baseball history to start the season, doesn't mean he's going to turn into an ace and throw a 1.50 ERA the rest of the way to finish the season at a 4.28 ERA. No, instead he's going to be that 4.28 ERA pitcher that Pecota predicted for the rest of the season, which then added with the six innings and 23 earned runs that already happened will mean that his final season line will be decidedly worst than a 4.28 ERA.

Again, a projection is just a projection. It gives you an idea of a player's or team's talent, but it is not exact. A team like the Marlins may be projected to 72 wins, but in actuality they may win 77 games with a Pythagorean record that suggested a 75 win team. That doesn't mean that 72 win projection was all that wrong, even though it was obviously not exactly correct. The Marlins may have been a 72 win talent team, but they got a bit lucky and bested their Pythagorean record by a few games, and even their Pythagorean record isn't an exact measure of how talented a team they are, just more accurate than their actual record.

So it's okay if projections are not perfect. But to think that a team that got off to a hot or slow start is going to automatically regress to their projection at the end of the season doesn't make sense.

The outcome of a game in April does not really affect the outcome of a game in July. It just doesn't. And until you can explain how the outcomes of games are dependent on each other, then the argument that a .444 team will likely revert to a .444 record at the end of the season even when they start out 11-3 doesn't make any sense either.

Apr 23, 2009 13:24 PM
rating: 0
 
jlefty

First of all, I don't see what good bringing up chien-ming wang does, considering we're already working with real world situations. I could just as easily bring up CC last year, who had a 7.88 ERA over his first 32 innings, and then a 1.95 over his next 221. Does this mean he was just off in his first 32 innings and that his true talent level when he is on is a 1.95 era? He's good, but most people would agree, probably not that good. Pitcher's are a whole different animal, lets stick to teams.

Read Joe's article today about strength of schedule. It's REAL simple. The SoS PECOTA used to make the .444 projection is very different from the SoS the Marlins have had so far. Pure and simple man. End of discussion....

But if you are willing to accept a projection, and than ONE game into the season, reject it...that strikes me as a little silly, no?

Apr 23, 2009 14:20 PM
rating: 0
 
Adam Hobson

A projection is a projection, whether that's for a team, a pitcher or a hitter. Their early season stats do still count.

I understand strength of schedule, however, even taking the strength of schedule into account, I doubt the 72 win projection accounted for an 11-3 start. It's not that pure and simple

On the other hand, I'm not actually the one rejecting the projection after one game or 14, that's you. If the Marlins are projected as a .444 team, then I figure that means they'll win .444% of their future games (adjusted for that strength of schedule of course).

You are the one who's decided that their initial 11-3 record means that the original .444 projection should now be reduced to a .412 projection for future games. I've asked this a few times before, but please explain how a win in April will cause a loss in July? How does a team that jumps off to a start well above their projection now cause that very same team to play to less than their projection the rest of the way? And strength of schedule explains only a fraction of that.

Apr 23, 2009 15:02 PM
rating: 0
 
jlefty

I keep dodging the question because its irrelevant. A win in april has nothing to do with july, obviously. But BP's projection was over a 162 schedule (based on a solitary number representing SoS.) You can't just split it into arbitrary sections and say they will be a .444 team in both sections. Otherwise, I could cut the season into 162 1-game sections and say that for each game I expect .444 wins, which rounds to 0, which means a .444 projection = 0-162.

A .444 projection doesn't anticipate an 11-3 start, or a 3-11 start, or any start. It anticipates a 72-90 finish. Again I ask you: Do you honestly believe projecting a .444 finish is equivalent to saying "They will definitely lose their first game."? Because your methodology is to take what they have already done, and then use the projection for the unplayed games. So a 72-90 projection is said to be incorrect if a team wins their first game. That is nuts. My method involves granting that BP's projection doesn't care how many games they win early on, only that they are only a good enough team to end the season with 72 wins.

Your method only works for coin flips, where the odds are STATIC. The odds of winning any given baseball game are not static, therefor, you cannot remove a .444 projection over 162 out of its context (i.e. the next 148 games).

Apr 23, 2009 15:28 PM
rating: 0
 
Ben Solow

Joe, you have a point in the sense that not all wins are equal, but Adam is fairly right about this -- if we assume that BP's projections perfectly represent the team's true ability, then there's not a reason to reject that it would represent our best guess of how a team will perform over a 148 game sample. While I haven't read Joe's article today, I would be absolutely STUNNED if the 14 games that the Marlins have played so far were easy enough in terms of SoS to make their SoS-adjusted PECOTA-projected winning percentage drop by a lot relative to the full 162.

You're also completely wrong in your conception of what it means to project a .444 winning percentage in statistics. Statisticians don't round expected values -- the expected value of one game in which you have a team with probability of winning = .444 is .444. The expected value of 162 games where a team has probability of winning = .444 is .444*162 ~ 72. The 72 is rounded at the END of the projection, not on a game-by-game basis.

The problem you're having is with time-inconsistency of projections....but this is how projections work. The 11-3 start is new information that was not available to BP when they made their original projection; they would likely come out with a different number if they did the PECOTA projections again using today's records as given. This is a pretty standard feature of dynamic programming with stochastic processes, and Adam's pretty much hit the nail on the head.

Apr 23, 2009 16:28 PM
rating: 1
 
jlefty

>>The 72 is rounded at the END of the projection, not on a game-by-game basis.

I know that. I know all about expected values, please, thank you. I was making a point that you can't arbitrarily cut the season into pieces and expect a .444 over each piece. Adam is expecting a .444 over the next 148 for the marlins. But if i cut the next 148 into 148 pieces, it doesn't work as shown above.

BP's projection was made within a certain context (i.e. 162 games to be played). One game of actual play doesn't change that, and you can't take the projection out of that context (if we are to assume it was a valid projection as you suggest). The Marlins have been getting pretty lucky with their wins so far. Let's assume that they are a .444 talent team over the next 148, like you say. That luck is still likely to catch up with them and they may start losing games they should have won.


"The 11-3 start is new information that was not available to BP when they made their original projection; they would likely come out with a different number if they did the PECOTA projections again using today's records as given."

The 1-0 start was new information too. You honestly think that BP's projection could not have possibly accounted for a 1-0 start??? because that's what you are signing up too.

BP thinks this is a team capable of winning only 72 games out of 162. Period. That they are on pace for more than that right now doesn't mean that BP MUST change its mind about its original projection.

Apr 23, 2009 19:04 PM
rating: 0
 
Adam Hobson

"Let's assume that they are a .444 talent team over the next 148, like you say. That luck is still likely to catch up with them and they may start losing games they should have won."

I think this is where the main confusion in your argument comes form. "Luck" doesn't actually catch up to a team. Even if a team was super lucky to start the season, that doesn't mean they are going to be super unlucky to end the season. The "luck" factor stays constant for each future independent game.

"So a 72-90 projection is said to be incorrect if a team wins their first game."

But I'm not saying that projection is incorrect right now. There is still a chance that the Marlins finish with a 72-90 record. However, what I'm saying is that after an 11-3 start, that projection is now less likely than a 77 and 85 record., if you still believe the Marlins to posses .444 talent. That's because the "luck" already happened to start the season, but that luck doesn't affect future luck at all.

Apr 23, 2009 19:38 PM
rating: 0
 
jlefty

>>But I'm not saying that projection is incorrect right now. There is still a chance that the Marlins finish with a 72-90 record. However, what I'm saying is that after an 11-3 start, that projection is now less likely than a 77 and 85 record., if you still believe the Marlins to posses .444 talent.

Be it an 11-3 start or a 1-0 start, you still think that the projection is no longer the most likely projection. After one game, that is absurd.

Apr 23, 2009 19:49 PM
rating: 0
 
Adam Hobson

Why is that absurd? When a projection is made prior to the start of the season, it's made as an educated guess. But once games have actually been played, then we have actual real information to rely on. When the 72-90 projection was made, we had no idea how any of the games would turn out. But now at this point of time we know exactly how 14 (actually 15) of the games turned out. We know that the Marlins won 11 of them. We know that they lost 3 (now 4) of them. We know that. Why should we ignore that new knowledge when we look at a projection for the the team's record at this point of time?

We know that the Marlins went 11 and 4 to start the season. However, if we still believe the Marlins have .444 talent, then that's the what we still expect them to do the rest of the way. We shouldn't ignore the new information.

Consider this, say at game 81, the Marlins are 40-41. Should we ignore those 81 games and still predict them a 72 win season? At game 162, if they are have an 81-80 record, we should throw that 72 game projection away, right?

Now obviously 81 games into the season, we may reevaluate the Marlins and decide, hey maybe they have .500 talent, not .444 talent. Or maybe we look at their Pythagorean record or whatever other statistics and decide, hey they really are a .444 talent team that got really lucky over the first half of the season. But still, regardless of what we think of the talent level at that point, we wouldn't throw away the results of the 81 games that already happened. Why now should we ignore the results of the first 15 games?

Apr 23, 2009 20:20 PM
rating: 0
 
jlefty

Honestly, I don't know what the number is. But I do know that it's not 1 game. I shouldn't have to say I was wrong about a projection after 1 game has been played. It's not nearly enough information to determine a talent level. And we still haven't settled the uneven schedule issue

Apr 23, 2009 20:50 PM
rating: 0
 
Ben Solow

Maybe we can get Nate to weigh in on this briefly? I mean, speaking as to my personal experience, revising expectations like this is practically the core argument of dynamic programming and the definition of Bayesian statistics. We DO have information now that we didn't have at the beginning of the season. What BP does is calculated one conditional expectation: E(wins | x1, ..., xN) where x1, ..., xN represent the set of information that we have about a team. What we're doing, after 14 games of the season, is calculating a different conditional expectation: E(wins | x1, ..., xN, and 11-3 in the first 14). There's no reason those two expectations should be the same....if the original projection is good the two conditional expectations should be close, but not the same unless the first conditional expectation had them winning 78.5% of their first 14 games.

Apr 23, 2009 22:37 PM
rating: 1
 
Adam Hobson

I'll agree with you that an uneven schedule does complicate things. Obviously if their first 14 games were against weaker competition than their remaining 148, I'd expect them to win a bit more often in their first 14 games. However, I think 11-3 is significantly better than a bit more often, so I think that 11-3 record still means they are beating their projection so far.

Apr 23, 2009 22:46 PM
rating: 0
 
Adam Hobson

Also, what's the record for the number of descendant comments? And do we eventually get to the point where it won't let us reply anymore, or will the comments eventually just be zero width?

Apr 23, 2009 20:24 PM
rating: 0
 
jlefty

honestly, at this point, finding out an answer to this is why I'm still going..

Apr 23, 2009 20:47 PM
rating: 0
 
Evan
(47)

Don't forget to adjust for strength of schedule. Half of the Marlins first 12 games were against the Nationals. Wouldn't most teams be expected to win at least five of those?

Apr 22, 2009 11:22 AM
rating: 1
 
dpbuckle
(867)

"The Marlins don't think that there is nothing unusual about their 11-3 record, the best in the major leagues."

Interesting sentence

Apr 22, 2009 10:34 AM
rating: 1
 
Pietaster07

The Marlins also won 2 of 3 againt the Mets and all three at Atlanta. The Pittsburgh losses seem to be a case of the Marlins not performing in cold weather and nothing worse than than.

Apr 22, 2009 11:54 AM
rating: 0
 
strupp

No love for Cubs @ Cards this weekend? Especially after mentioning the Bradley incident and the Lou-TLR barbs?

More east coast bias :)

Apr 22, 2009 14:29 PM
rating: -1
 
You must be a Premium subscriber to post a comment.
Not a subscriber? Sign up today!
<< Previous Article
Premium Article Future Shock: Further ... (04/21)
<< Previous Column
Premium Article On the Beat: Cooper's ... (04/19)
Next Column >>
Premium Article On the Beat: Setbacks ... (04/26)
Next Article >>
Prospectus Hit and Run... (04/22)

RECENTLY AT BASEBALL PROSPECTUS
Premium Article League Preview Series
Every Team's Moneyball: Minnesota Twins: Reb...
Premium Article Skewed Left: History Repeats Itself
Premium Article League Preview Series
Premium Article Pitching Backward: Why Relievers Get A Free ...
Premium Article Spring Training Notebook: Cactus League
Prospectus Feature: How the Astros do Spring...

MORE FROM APRIL 22, 2009
Premium Article You Could Look It Up: Reality Therapy
Premium Article Under The Knife: Mid-Week Update
Premium Article Ch-Ch-Changes
Prospectus Hit and Run: Tales in a Minor Key

MORE BY JOHN PERROTTO
2009-05-03 - Premium Article On the Beat: Outside and Looking In
2009-04-29 - Premium Article On the Beat: North Country Thunder
2009-04-26 - Premium Article On the Beat: Setbacks on the Eastern Front
2009-04-22 - Premium Article On the Beat: Reeling in the Marlins
2009-04-19 - Premium Article On the Beat: Cooper's Optimism
2009-04-15 - Premium Article On the Beat: Cardinal Creativity
2009-04-12 - Premium Article On the Beat: Week One Wrap
More...

MORE ON THE BEAT
2009-05-03 - Premium Article On the Beat: Outside and Looking In
2009-04-29 - Premium Article On the Beat: North Country Thunder
2009-04-26 - Premium Article On the Beat: Setbacks on the Eastern Front
2009-04-22 - Premium Article On the Beat: Reeling in the Marlins
2009-04-19 - Premium Article On the Beat: Cooper's Optimism
2009-04-15 - Premium Article On the Beat: Cardinal Creativity
2009-04-12 - Premium Article On the Beat: Week One Wrap
More...