Are the White Sox PECOTA’s Computer Virus?

Well, it’s that time of year where PECOTA predictions come out and doom is the forecast for the upcoming White Sox season. Since the start of Ozzie Guillen’s managing career, PECOTA has predicted win totals of 79 (2004), 80 (2005), and 82 (2006). The actual season-ending win totals have been 83, 99, and 90, respectively. So in the 3-year managing career of Guillen, PECOTA has been off a total of 31 games. I’m not going to look through every other team during this period, but I’ve got to guess the White Sox have made the system look worse than any other team.

So is it any wonder that White Sox GM Kenny Williams would respond with the following quote, when asked by the Chicago Tribune’s Dave Van Dyck what he thought about the PECOTA prediction of 72 wins in 2007?

That’s a good sign for us because usually they’re wrong about everything regarding our dealings,” general manager Ken Williams said. “What can you do? We put the best team together we can, and we think we’re going to end up somewhere in the mid-90s, although there are all kinds of variables off that.

Now, I don’t believe that PECOTA’s creator (Nate Silver) is some diabolical genius with an axe to grind against the White Sox, but I have questioned his findings in the past, when it comes to the South Siders. 72 wins for this season is the most ridiculous of all of them.

Let’s focus on the starting pitching, as this is where PECOTA usually is the farthest off on, when it comes to the White Sox. I strongly disagreed with PECOTA in 2005 and even I couldn’t have guessed how far off it would end up being. I’m getting the same feeling about the forecast for 2007.

NAME ERA WHIP
Buehrle 4.90 1.43
Garland 4.83 1.43
Contreras 5.15 1.46
Vazquez 4.45 1.31

While these pitchers are hurt by throwing in park favorable to hitters, the PECOTA findings are staggeringly bad for the top 3 pitchers in their rotation.

Contreras is supposed to go from an ERA of 3.61 (2005) and 4.27 (2006) to one this season well over 5. His WHIP is projected to rise .21 from what he has averaged the past 2 seasons.

The statistical peak age of baseball players is generally considered between 27 and 28, which are the ages of Jon Garland and Mark Buehrle, respectively. Despite having an ERA of .3.50 (2005) and 4.51 (2006), PECOTA says Garland will drop 1.33 from what he did when he was 25. Buehrle, going into his free agent year is expected to finish with an ERA over a run more than his career average. Neither of these pitchers have ever had serious arm issues, but they are both seen by PECOTA to be falling off the cliff.

Few starting staffs were pushed harder than the White Sox of 2005. While these pitchers looked tired during 2006, Guillen has always done a great job of staying away from putting them in PAP situations. No team over the past 5 years has had fewer injuries than the White Sox, which does has some luck behind it, but more importantly speaks to how the organization selects players and how great their training staff is.

My projections for the White Sox pitching staff for 2007 are the following.

NAME ERA WHIP
Buehrle 4.45 1.34
Garland 4.15 1.29
Contreras 4.40 1.35
Vazquez 4.25 1.31

While I do believe that Thome, Konerko, and Dye will be off their torrid hitting of 2006, I still think all of them will exceed their PECOTA projections of 2007. I think Uribe/Cintron at shortstop will be much improved over 2006, while the addition of Toby Hall will add more to the catching department.

The big question mark is the possibility of an Erstad and Podsednik tandem, killing the leadoff spots of the batting order. I fear this as well, but I’m hoping that sanity will take over and Brian Anderson will play most of the games in center. I think Anderson will show the biggest improvement in OPS of any regular from 2006 to 2007, if given the chance. Considering what a weak spot left field was for the Sox last season, Erstad and Podsednik splitting most of their bats in 2007 I think will actually be a slight improvement. A better option would be to put prospect Josh Fields in the mix, but it appears the White Sox don’t want to add service time and plan to give him the starting 3b job, when Crede leaves for free-agency after this season.

Where the big improvement in the team for 2007 will be is in the bullpen. After being great in 2005, the pen has been almost completely rebuilt with Cotts, Hemanson, Politte, Marte, and Vizcaino gone and hard-throwing Thornton, MacDougal, Massett, Aardsma, and Sisco bringing a whole new look to the White Sox.. The one bullpen holdover from the championship season, Bobby Jenks, will close out the games again. Jenks is the one player on the team I worry about, but the Sox have plenty of options to replace him, if he physically or mentally implodes.

After beating the Pythagorean odds for the past 2 seasons, I do think the Sox will have a hard time doing it for a third time, but this year’s model is not a team which will win only 72 games. The AL Central might not have the 2007 World Series Champ, but its top 4 teams have a good chance of winning at least 82 games.

Unlike their competitors, the White Sox seem to be in the best shape to add a contract or two around the trade deadline, which is a direct benefit in trading Freddy Garcia, who I suspect will have a higher ERA or WHIP than any starter the Sox have on their staff this season. In what I expect will be a very tight race, this could be the thing that pushes the White Sox over the top.

Let me finish by mentioning that I’m not knocking PECOTA overall. I’ve seen the data and it has done best of all the systems that predict player performance. I have owned every Baseball Prospectus annual since discovering it in 2003, so I’m a big fan of the website. As you might know, one of my best friends just happens to be a major attraction for BP. I’ve even appeared as a guest on Baseball Prospectus Radio a couple of times.

Having put out these proclamations; let me say that I do think that there is an overall bias against the White Sox by the staff of BP. Michael Lewis’ “Moneyball” made Kenny Williams appear to be a dunce to the brilliant machinations of sabermetrican’s golden boy, Billy Beane. An underlying subtext not discussed by Lewis was that the White Sox use a mix of stats/scouting. (I don’t blame Lewis for framing it this way. He was writing a story and it was a great one.) Like most in the sabermetrical field, BP has to have a rooting interest against the White Sox, as the brash Williams has taken shots at the sabermetrical followers, even though he and his staff consider statistical factors in their decision-making.

At least once a year, I defend Williams for his moves, as I think he is one of the best GM’s in baseball. I would think that loyal readers here by now realize that I don’t aspire to come off like a homer (I’m a White Sox fan), as I prefer the contrarian path. Having Jerry Reinsdorf as owner, Williams as GM, Guillen as Manager, and Hawk Harrelson as the voice of the team, doesn’t exactly come off collectively as a group worried about being media friendly. While Harrelson is an idiot when it comes to his thinking about OBP and other statistical measures, I think the other 3 play-up their Ozzie-ball style, while actually understanding how a sabermetrical approach has its virtues.

During the period between 2000-2006 there have only been 5 teams who have not had a losing season; The Yankees, Red Sox, A’s, Cardinals, and White Sox. I don’t see where in 2007 it will mark the end of this streak for Chicago.

19 thoughts on “Are the White Sox PECOTA’s Computer Virus?

  1. 1.  Nate Silver, from BP Unfiltered last friday:

    “I’ve publicly disavowed PECOTA’s projection that the 2005 champs will finish with 90 losses, but it wouldn’t surprise me one bit if that’s where the 2006 champs wind up.”

    Joe Sheehan, July 6, 2006:

    “With Guillen one of the few managers who actually brings something to the table, and Kenny Williams’ development from a bumpy start to become one of the better GMs in the game, the White Sox have one of the best management teams in baseball. There’s absolutely no way I would have expected that two years ago, and quite frankly, there’s probably a terrific book in examining how these two men have ascended to the top of their profession. Williams’ career path–from Todd Ritchie to Jim Thome–is one of the most fascinating I can think of for an executive.

    Moneyball II, anyone?”

    I still haven’t figured out where exactly Silver has “publicly” (ie in print) disavowed the White Sox projection, but I’m willing to take his word for it. If you’ve read Sheehan regularly the last few years, he’s been very clear about how his opinion of Kenny Williams and the White Sox has changed, although whether this is unique to Joe or a more general BP trend, I couldn’t really say. Just wanted to get a little more primary source info into the discussion.

  2. 2.  Thanks for the quotes. I have read themm since I read all of Sheehan’s and most of Silver’s articles. I appreciate the evolution of Sheehan on the subject. Prior to the 2005 season, Joe was predicting a similar record 70+ season for the White Sox.

    I have picked the White Sox to finish 2nd the past 2 years, so I’m not saying they were a slam-dunk choice, but the general consensus at BP has rated them below what the major media publications have both years.

    As I wrote in the piece, I’m not saying Nate Silver’s PECOTA projection was somehow corrupted by a bias he has against the team. My point is that the White Sox seem to be the kryptonite to his system. Because PECOTA is such a strong feature of BP, I have to guess at this point that most at the site root against the White Sox, as the team’s management has publicly scoffed at the prediction tool, making PECOTA look badly in the public eye (see Tribune story).

    I know if PECOTA was the number 1 product at my company, I would be rooting against the White Sox. Silver lives pretty close to US Cellular, so don’t be surprised to see him at games this year with voodoo dolls in hand.

  3. 3.  I think PECOTA has its strong points – forecasting the range of performance by individual players. But while in theory the sum of the players performance equals the team’s performance, trying to predict that leads to problems. The manager gets to decide who gets playing time and when. While two teams may have all their players hit their 50% PECOTA OPS, if one manager maximizes the playing time of his best players, the team will perform better than predicted. The ability to replace injuries, add to the team depth, all affect the team’s performance without changing anyone’s underlying PECOTA forecast. SO if a management team can out manage and out GM the other teams in the league, they will out perform the team PECOTA projection.

    I don’t think the measure of a good manager is the team outperforming the end of year Pythangenport wins and losses, but getting the most out of his players who are outperforming their PECOTAs.

  4. 4.  2 – See, I don’t think Nate would be there with a voodoo doll, in fact, I would think it would be the exact opposite. He realizes (as does anyone else actually doing projections) that you’re never going to get much better than 75% accurate or so with any projection system, and the trick is not only coming up with the projections, but being able to figure out where the gaps are and to account for them. BP’s business strategy should not be “Convince everyone that PECOTA is always right” but rather “Convince everyone that PECOTA is the most accurate, while being able to explain to them where it might be making mistakes.” As nice as it is to be able to call it “deadly accurate” and cite a few good examples on the cover of the book each year, it’s really about coming up with the best baseline you can before adding in whatever intangible or otherwise unquantifiable information that can make the analysis of a player or team even better. While PECOTA may be BP’s #1 “tool,” the most important “thing” from a business standpoint is really the writing and writers themselves. PECOTA just gives them a starting point.

    And another important note is that the PECOTA team projections are one place where at least a fair amount of “guesswork” is added in, namely, in terms of playing time predictions, which are notoriously difficult. The individual player projections are almost entirely data-based, but the team ones need a little more tweaking. Hence the occasionally screwing results.

  5. 5.  I hear you, but people are not robots, as rational and logical as even a sabermetrician can be. If someone (Williams and the White Sox) is out there taunting the performance of your product, it’s next to impossible not to root against them. I know the chip I permanently wear on my shoulder would behave this way, so I have to believe that even more enlightened people than myself couldn’t completely supress this urge.

    The biggest point of the article is that since Guillen took over, PECOTA has underrated the White Sox.

  6. 6.  5 – I had actually initially written something addressing the fact that having a bias against the White Sox/Williams for calling them out would be understandable, albeit not the most “rational” thing in the world, but I ended up re-writing the comment. But I also think that Nate would probably have preferred to get out in front of the “prediction” story, so that it was more clear that while PECOTA may have come up with a strange result, he didn’t agree with it. Baseball Prospectus didn’t pick the White Sox to lose 90 games, PECOTA did, and it’s up to Silver and the rest of the BP staff to explain the difference. While stirring up a fight with the White Sox front office may be a valid PR strategy, I’d be surprised if it was one that they followed deliberately.

  7. 7.  But yeah, I’d be curious to know if there’s a systemic reason that PECOTA has been consistently bearish on the Sox. I think part of the explanation might be in the White Sox chapter of the 2006 book, where Silver unveils the aptly named but as-of-now unpopular “Persistent Underlying Team Ability” system, but I know that’s not the whole story.

  8. 8.  I don’t know about past years, but IIRC, this year the primary reason for PECOTA being down on the White Sox is the age of many of their hitters.

    I do know I’ve read many BP articles where they extol the virtues of Ozzie Guillen’s handling of his bullpen and rotation, and his roster in general. I don’t have time to find cites to specific articles right now, I’m afraid.

    Of course, then I read something like this

    http://tinyurl.com/2how76

    and while I don’t agree with everything the FJM guys say, you’ve got to wonder just how well a lineup with Pods and Erstad in front of the big guys (ie, Konerko, Dye, and Crede) will affect things – as Scott himself worries about above.

  9. 9.  8 – To save you a little time, I’ll just copy the paragraph just before the one I quoted earlier from the Sheehan article:

    “If this looks strange to you, I assure you that you’re still reading Baseball Prospectus. It may be hard to remember, but we were praising Guillen back in 2004 for his handling of the starting rotation, and last year for his use of the bullpen. Three years into his tenure, it’s become clear that Guillen is one of the best managers of pitching in baseball. Forget smallball; it’s the pitching where Guillen has made his greatest impact on the Sox.”

  10. 10.  I’ve read these articles as well, but when it comes to where the writers for BP pick the White Sox to finish, they for the most part pick the White Sox lower than most major media does. While most saw the White Sox finishing 2nd in 2005, the majority of writers at BP had them finishing 4th. I will be surprised if this changes in 2007, as just not at BP, but most in the sabermetrical community dislikes the White Sox.

    Let me state once again that I love Baseball Prospectus. Besides Ken Rosenthal and the Toaster, it’s my favorite place to get baseball information. Just last week I mentioned I thought Sporting News should hire Joe Sheehan to be their baseball writer. My piece is based on the idea that the White Sox have defied BP predictions and that PECOTA seems to be off on the White Sox during the Guillen era.

  11. 11.  10 – Well, last year, PECOTA had them winning 82 games and finishing 4th, in a tight race. (Remember, PECOTA regresses, so it’s hard for any team to be picked to win 95 and run away with it). The Staff picks, however, were much more favorable, with the White Sox finishing with an average rank of 1.75, just behind Cleveland (1.58). Only one of the 12 writers, Jonah Keri, picked them to finish even 3rd, which is where they ended up. So, since you haven’t enjoyed our dancing around and talking about other interesting and related issues, I’m happy to bring things back around to your point.

    http://www.baseballprospectus.com/article.php?articleid=4921

  12. 12.  Nate Silver begins his series on how PECOTA predictions jibe with his own thinking. SHould be interesting to see his Central preview, considering the White Sox and Cardinals have not had a losing season this century, but are seen to be going that direction this year.

    ONCE AGAIN LET ME STATE: I love BP, just think the White Sox are a team they don’t generally have a good handle on.

  13. 13.  12 – Scott, apparently you think that the reason that I’ve ignored your continued statements about how much you like BP is that I don’t believe you, and that my comments have been an attempt to show that you are lying. This could not be further from the truth. I know that you like BP and aren’t trying to say otherwise. But you still haven’t explained how the entire BP staff picking the Sox to finish 1st or 2nd in the division last season showed that they “don’t have a good handle on them,” in the sense that they underestimate them. I certainly see your point about PECOTA itself, and I’m pretty sure that in 7 I clearly stated my interest in finding out if there is a systemic reason. Nate’s comments today on the Phillies (a team that you can easily make a case for as being the White Sox unlucky dopplegangers) suggest that such an answer may not be forthcoming, but it still doesn’t change the fact that I’m not convinced that the PEOPLE at BP have the level of bias or underestimation that you are ascribing to them.

    And once again, yes, I know that you love BP, and am not and have not tried to say anything to the contrary.

  14. 14.  Scott–

    Maybe do a “Please Explain: Chicago White Sox” and then have Will make sure everyone at BP sees it.

    As for pitchers, PECOTA (that P stands for Plexiglass, doesn’t it?) always seems to squeeze folks towards the middle: check even the predictions for Johan Santana and Chris Carpenter.

  15. 15.  As I wrote in the piece, I’m not saying Nate Silver’s PECOTA projection was somehow corrupted by a bias he has against the team. My point is that the White Sox seem to be the kryptonite to his system.

    Maybe you can get Nate to research his own system to determine potential factors that PECOTA overlooks to explain why it keeps missing with the White Sox.

    Personally, I find frustrating that I’m not aware of any studies published by the BP crew on the historical accuracy of PECOTA, when it misses, what it is most accurate at doing and what it is least accurate at doing. It seems to me that researchers should publish findings for the scrutiny of the larger community, but BP has a financial disincentive to do so.

  16. 16.  Someone might say that I refused to respond to your point, Ali, because it would take away from my argument. I would be that someone. I did notice what you wrote when researching for my piece. While I could mention that most of the mainstream media had the White Sox in first, while more BP writers had them predicted 2nd—it’s basically splitting hairs and I was with BP in thinking that Cleveland would win the Central in 2006. (Maybe I should stick to picking football games considering this and my NCAA tourney brackets over the past year.)

    So I have been motivated by you and George’s funny point to follow-up on my piece.

  17. 17.  One thing. I seem to recall in 2005 that PECOTA projected the White Sox to win 82 games. Then Joe Sheehan said that since the White Sox always underperform their projections they’d win only 71. That didn’t turn out so well. Beware the reverse effect. PECOTA has a built in bias against pitchers who don’t get strikeouts, which describes all the Sox pitchers besides Vasquez. Anyway, I put money that the Sox wouldn’t finish 9 games better than the Devil Rays based on that pick so I’m hoping it’s true.

  18. 18.  When the White Sox were built to slug the ball, not focused on defensive skills and had a shallow bullpen, they would underperform their PECOTA.

    See Cleveland Indians for another team that now fits that profile.

Comments are closed.