Keith Woolner is an author of Baseball Prospectus.
Keith Woolner: Hi everyone, let's get chatting...
Peter (NY,NY): Hey Keith, who are you voting for in this 'final vote' thing.My picks are Frank Thomas and Bobby Abreu (when is that guy going to get his due?).
Keith Woolner: Thomas and Abreu would probably be my picks, especially since I consider established track record versus current season stats more than most.
Hraboskyfan (Chattanooga, TN): What's the best way to compare pitchers and position players in terms of overall value?
Keith Woolner: In terms of measuring past run production and prevention, something like VORP does this (especially if you combine VORP with a measure of defense at the position).
It becomes more complicated if you talk about future value. While a pitcher and hitter may have the same expected value going forward, the pitcher probably has a higher injury risk. Depending on what you're valuing, you may discount his future contribution accordingly.
Johnny Damon (Boston): Keith, how much would you pay for a lock of my hair? I'm considering cutting it for charity, too, after shaving my beard was so successful.
Keith Woolner: Given the rate at which my own hair is disappearing, I'd probably be willing to pay just under the going rate for Rogaine.
Brian Durack (Texas): Are there any good studies available that show how walks translate from the minor leagues (specifically AAA) to the majors?
Keith Woolner: Clay Davenport has an study that compares translated rates between the Davenport translations and MLE's. The link is:
Note though, that the study is a few years old, and doesn't reflect updates to his methods.
Terrence (Andover, MA): Who gets the wild card? Sox over Texas?
Keith Woolner: I'll go with the Red Sox, both for the irrational reason that I am a Bosox fan myself, and the (somewhat) more rational reason that I don't think Derek Lowe is this bad, nor Kenny Rogers and Ryan Drese this good.
Randolph (San Ramon, CA): What's the best way to measure the defense of a team? DER seems to end up being almost entirely dependent on strikeouts, and it seems like it'd be highly dependent on park. Is there ever going to be a way to separate pitching from defense, and effectively measure the relative value of defense performance from player to player?
Keith Woolner: DER is probably more dependent on the team's groundball to flyball ratio than strikeouts, but your point still stands.
Within the past few years there's been an immense amount of new kinds of data collected -- better precision on the location of batted balls, velocity, hang time, position of fielders, etc. This data, in principle, gets us much closer to isolating individual player defensive performance.
What hasn't happened yet is a clear, consistent, and well-accepted way of interpreting that information into a metric. Part of the problem is simply the volume of data itself -- very few organizations (and/or aspiring analysts) have both the computing horsepower and the statistical sophistication to make good use of it.
Park factors clearly will need to be part of the solution though. It's even possible for instance, that parks will play differently between day and night games due to the position of the sun and how the outfielders react to it.
Kas Gardiner (St Louis): Why don't the A's give Justin Duchscherer a shot in the rotation. He was lights out early in the year (granted, in low leverage situations), but his numbers in AAA were BP friendly. Yet he gets no "love," not even from BP. Is he a prospect and if not, what's he lack?
Keith Woolner: I'm not sure a 26-year old pitcher deserves the "prospect" label much longer, but the A's still have plenty of love for Justin. You're correct that he's been pretty good according to BP's translated stats. I suspect that he'd get more attention if he:
1) Had a great fastball
2) Threw the knuckleball
3) Pitched submarine-style
wmcdonal56 (notre02): Would you rather be a Royals fan or a Tigers fan for the next decade?
Keith Woolner: Royals, if only so I could hang out more with Rob and Rany (http://www.robneyer.com/robrany.html)
Darren (Vancouver, BC): Hello Keith. I'm a big fan of the stat reports. Thanks for doing such a great job. My question is how would I go about calculating state tables for base out conditions for individuals. It seems like managers should have those in front of them.
Keith Woolner: The problem with computing individual hitter situational tables from the actual game data is that hitters don't see enough plate appearances in different base-out situations to assess the expectation properly.
Even if a batter with 600 PA had those chances perfectly distributed across each of the base-out states, you're still talking about a sample size of 30 or less for each case.
In this case, a simulated approach would probably be a better way to go.
Mark (Long Beach): Thanks for taking my questions. Do you see Brad Lidge as a future dominant closer? Also, do you think the A's should have gone after Lidge instead of Dotel (assuming Houston would have parted with either)?
Keith Woolner: I think that just about any reliever who has a track record of dominating in the majors like Lidge has could be a successful closer if given the chance.
JC (Manhattan): Kris Bensen has done nothing for a while now...can he be the next Jason Schmidt?
Keith Woolner: There are some intriguing similarities between Schmidt and Benson -- both became fulltime major league starters at age 24, both suffered through arm problems around age 26-27.
However, where they differ rather strongly is that Schmidt's strikeout rate spiked upwards almost immediately after he came back -- jumping from 5.7 K/9 to 7+ K/9. Benson hasn't been able to show that kind of dominance yet, so predicting him to break out and become one of the league's best pitchers is more wishcasting than inevitable outcome.
Alberto Leiter (NYC): Does the Mets' sweep of the almighty Skanks...oops...I mean Yanks mean anything?
Does it make the Metsies a better team than they were before the weekend series?
Keith Woolner: Well sure, they're 3 wins better than they were before, right?
A less smart-aleck and more stathead-ish response would be that those three games represent when combined with their previous performance, a further sample of what the Mets actual "true level" of ability is. So they may be slightly better than we thought they were, but only a little.
That's unlikely to become their marketing slogan, though.
Rufus (Arlington): Would you prefer the All-Star managers to try their hardest to win, or do you enjoy seeing all the players?
Keith Woolner: I take a rather extreme position on the All Star Game. I couldn't care less about the outcome of the game itself. I consider it an in-season encore for the game's best, and as such ensuring that everyone get's into the game to be recognized is essential.
In fact, I would favor a free substitution rule that allows players who were taken out of the game earlier to appear again, ensuring that every position player has appeared by, say, the 6th inning.
I'm also a fan of the "at least one player from every team" rule, and would expand the rosters to 40 players to take into account the growth in the size of the leagues over the decades.
Murder, Inc. (Old Chicago): How about some more graphics on the site, particularly with regards to the stat reports? A histogram of MLVr would be cool by position, you could automate it, and broaden the appeal of your stats beyond us cool math dudes.
Keith Woolner: Excellent idea, and one we've been talking about. As I've alluded to in previous chats, we've done a lot of work this year behind the scenes to automate and consolidate a lot of the stat reports and other data-driven features on the site (e.g. the Team Tracker under BP Fantasy). It's taken longer than I had hoped for to get the technical infrastructure into place, but we're very close to being able to develop some very, very cool graphical stat presentations.
R.J. (Washington, D.C.): Freddy Garcia for 3 years, $27 million? Kenny Williams seems to think "aggressive" means "smart." Tell me this isn't a totally idiotic deal.
Keith Woolner: That's way overpriced for the market, of course, but if it makes you feel any better, Garcia's PECOTA comparables heading into 2004 included John Smoltz, Jack Morris, and Jim Lonborg, which represents a pretty decent upside. His rebound in 2004 looks to be at about his 80% percentile PECOTA projection, which is also comforting.
(Granted, those three weren't his *top* comparables, but they are on the list...)
Josef (Wilkes-Barre): Do you think all the win modeling that's done using baseball analysis really makes sense? You've got conflagrating issues with the models of runs scored, including components with varying leverage as static, like stolen bases. Then, it's all plugged into a simple RS^n/(RS^n+RA^n) model to calculate winning percentage, and then your financial guys use marginal $ gain per win to calculate value. Aren't there enough error bars in that system to make the whole exercise nothing more than a coarse guess?
Keith Woolner: Sure, the error bars do get wider when you build more complex models out of simpler ones, but even the simple ones are only an approximation of reality.
The sabermetric community has never been good at acknowledging the error bars, in part because it's difficult to communicate effectively to a larger audience used to concrete answers. PECOTA, though, tries to accomodate this through expressing forecasts within a probability range.
In any event, the real value for a model is whether it provides useful information for decision making, and whether it does so better than comparable alternatives for that information.
Keith Woolner: Sorry gang, but that's all the time we've got tonight! Thanks for all the questions, and for tuning in. I think Jim Baker is next up in the chat rotation, so look for him here on Thursday!