Sunday, March 25, 2007

Some Surprising PECOTA Team Projections

Baseball Prospectus released their annual PECOTA projected team standings a little early this year. Though the methodology is certainly sound and the results nearly as accurate as possible, the exercise never fails to deliver a few W-L forecasts that turn heads. One of these, the 72-90 projection for a team that's won 189 games over the past two years, has even found its way into the clubhouse, not that they've let it get them down.

In this entry, we'll look at a few teams whose PECOTA output differs significantly from their public perception (as measured by the Las Vegas over/under line) to see if we can shed some light on the gap. We may as well begin with the aforementioned...

Chicago White Sox

Vegas line: 87.5 wins

PECOTA forecast: 72 wins

To a typical baseball fan, projecting a team to drop 18 games in the standings from 2006 and 27 from 2005 may seem like sheer lunacy. Of course, that fan is living in the era of ESPN-driven hype, where it's easy to see this year's Pale Hose combining 2005's excellent pitching with 2006's offensive output, but not the other way around. As it turns out, PECOTA is not the only projection system that foresees a sub-.500 season on the South Side; all the major ones do.

The Sox do have a confluence of factors working against them, something I've written about before. In short, they received a lot of surprising offensive performances last year, stayed impeccably healthy, and their pitching staff had a collective career year in 2005. All these things are likely to regress this year, although some certainly can't believe this is the case. Despite the number of recognizable faces on the team, the Sox have no true superstars; no one is projected to top 30 VORP in 2007.

At the same time, I think they will win more than 72 games. Yes, they play in a tough division and a tough league, and yes, they will regress substantially, but the current BP team depth chart sees 498 PA for Darin Erstad (compared to 505 for Jim Thome) and 100 innings for Gavin Floyd, neither of which seems likely. Just balancing out the team playing time distribution adds a couple of wins to the forecast, bringing us up to 74.

Though no individual player's projected line seems off by a huge amount once you account for Thome's age and Jermaine Dye's lack of a track record, the collective bearishness of PECOTA on the offense is a little hard to swallow. Though the computer is much better at this than I am, I'm inclined to add two wins on this basis. The pitching forecasts look about right, so no change for those. We're now at 76.

Some have argued that the team is in a good position to make a big midseason acquisition, and should receive bonus wins for this in their projection. Before this past offseason, I might have bought this, but it seems that GM Kenny Williams is now more interested in taking advantage of a seller's market, and he has several desirable players, including Dye and Mark Buehrle, whose contracts expire after 2007. What's more, the Sox aren't particularly likely to be leading the division or wild card race in July, and have a shallow farm system, making a blockbuster deal unlikely. A top midseason acquisition will likely only add a win or two to the equation anyway. No change for this factor.

PECOTA also can't account for skilled managerial or medical staffs, but it's difficult to measure their benefit, and thus difficult to account for them in a team forecast. I'll add one more win, but this effect could easily be off in either direction.

Final Verdict: 77 wins

Tampa Bay Devil Rays

Vegas line: 67 wins

PECOTA forecast: 79 wins

Nate Silver, the mastermind behind PECOTA, has already done this little exercise for the NL and AL East, and concludes that the Devil Rays should lose about three wins from their forecast.

What's interesting here is that while the other projection systems agree that the White Sox will not top 80 wins, no one else sees the Devil Rays straying significantly from their 67-win Vegas line, and certainly they don't anticipate them finishing seven games above the 2005 champs.

I can't fully explain the difference between PECOTA and the other systems, although certainly the team ERA is significantly lower in PECOTA. Since pitching forecasts are inherently less accurate, I'm inclined to deduct two wins for the outlier effect. Running total of 77.

As Silver points out, Rocco Baldelli is reasonably likely to be dealt for pitching prospects at some point during the year. If the team actually does make a run at .500, it will be interesting to see whether the front office keeps the faces of the franchise around as a PR move. The Devil Rays have a surplus of young outfielders, but will need more good young pitching to compete in 2009-11 and are unlikely to get it without dealing an established regular or two. Combining the trade possibilities with an aggressive projection for Akinori Iwamura and a gaggle of insufficient first base stopgaps, I'm deducting two wins from the offense.

Final verdict: 75 wins

Arizona Diamondbacks

Vegas line: 78.5 wins

PECOTA forecast: 88 wins

At least in this case it's easy to explain the gap of 9.5 wins: the 2007 D'backs will feature a ton of rookies, and the linesmakers and public fail to appreciate just how good they are. Digging deeper, they have only one real star, Brandon Webb, who is underappreciated by the public. Nobody thinks Randy Johnson has anything left, but they will likely be surprised. Chris Young is projected to instantly become a monster in center field, and Stephen Drew and Orlando Hudson should rival Rafael Furcal and Jeff Kent as the league's most valuable keystone combo west of Philadelphia.

What adjustments need to be made? Hudson is one of my favorite players, but asking him to set a new career high in VORP seems a little much. Meanwhile, I don't think even Young's parents expect him to top a .900 OPS this year, and Chad Tracy has been the subject of frequent trade discussions. Two wins get lopped off for the offense.

On the pitching front, 120 innings for Johnson may actually be a conservative forecast, and I don't think his 3.78 ERA is unreasonable at all given a return to the NL and his solid peripherals in New York. Livan Hernandez has defied PECOTA for years and I think he should be worth five runs more than his projection. I do wonder if Edgar Gonzalez can maintain a sub-5.00 ERA in the bigs, but he can be replaced by Enrique Gonzalez or Dustin Nippert with no real loss.

While some say Jose Valverde needs to get a better head on his shoulders or become more consistent, I point to his excellent K rates throughout the past four years and the inherent variance involved in 50-inning relief "seasons," where one cold spell can kill your bottom line. He should match or beat his 3.71 ERA. Jorge Julio's name pops up in trade talks, but if he is dealt it will be for something of immediate and comparable value. Overall, the pitching staff is worth one more win than advertised.

Final verdict: 87 wins


As I mentioned earlier, Nate Silver is tweaking his projections on his own. While I'd be interested in seeing whether this improves the forecasts, there's really no realistic way to filter out the "noise" in a 30-team sample. Plus, according to Diamond Mind's three year prediction rankings, it seems Silver should stick to what his computer tells him. After all, it is "deadly accurate."

No comments: