Baseball Prospectus provides what is probably the most famous baseball preseason predictions. Since actual games are not yet being played, I have been taking a deeper dive into the projections.Recently I posted a criticism of Baseball Prospectus' prediction for the 2010 Yankees. That piece ran before BP updated its projections on Monday. For those disinclined to read either post again, BP essentially predicts that many key Yankees will give way to age, even though the team is still projected to finish atop MLB. My criticism of the BP projections can be best summed up below:
Derek Jeter - 2009 actual VORP: 72.8, 2010 projected VORP: 30.3
Jorge Posada - 2009 actual VORP: 35.7, 2010 projected VORP: 16
Mariano Rivera - 2009 actual VORP: 29.5, 2010 projected VORP: 14.3
Phil Hughes - 2009 actual VORP: 25.6, 2010 projected VORP: 10.9
Until the recent change, I would have also included A.J. Burnett and Andy Pettitte on this list, but their new projections are modest regressions from last season. That improvement aside, BP projects that the four players above will attribute for a loss of 92.1 runs from their 2009 performance. Each one is projected to lose more than 1/2 his value.
A rule of thumb is that 10 runs is about 1 win. Meaning ... the difference between CAIRO's 102 win projection for the Bombers and BP's updated 94 win projection is attributed almost entirely to those 4 players (nevermind some of the other questionable declines PECOTA forecasts).
I have no idea what PECOTA has against Phil Hughes, but my argument has been that since I've been paying attention to it, PECOTA has struggled to project truly outstanding production and quality production from players who contribute as they get older. However, if that observation is to have any merit it needs to hold for the rest of baseball, given that I obviously pay disproportionate attention to the Yankees, and the Yankees have a lot of future HOFers and aging superstars. For the purposes of brevity and local rivalries, in this post the part of the rest of baseball will be played by the Boston Red Sox.
Given that our legal budget is spread thinly as it is, I will avoid posting the actual predictions because Baseball Prospectus charges for them. However, based on my quick, back of the envelope calculations, my criticism stands. Unlike its projections for the Yankees, PECOTA doesn't have as severe a season-to-season volatility in its projections for the Red Sox.
That is not to say that PECOTA is putting forward credible estimates at each position. The following 2 predictions stand out for being bad:Kevin Youkilis - 2010 projection: .288/.387/.500, 28.7 VORP; 2009 actual: .305/.413/.538, 53.5 VORP. I don't like Kevin Youkilis either, but why would he lose almost half his value and post his lowest OBP since 2006?
Victor Martinez - 2010 projected VORP: 23.9; 2009 actual VORP, JUST WITH THE RED SOX: 21.4. Granted, V-Mart overachieved once he got to Boston, but a full year of his production should absolutely be better than his partial year in Boston in 2009.
But that's it on offense. There are some players, such as J.D. Drew and Jacoby Ellsbury, who are projected to see a sharp decline in VORP, but whose raw performance projections from which VORP is calculated seem reasonable. VORP is a counting statistic, a relative statistic, and inherently flawed, so I will avoid criticizing a drop in projected relative value if I feel the performance estimate is accurate.
On the flip side, while virtually every Yankee is projected to LOSE VALUE, many Red Sox are projected to improve. Let's choose one, entirely at random:
David Ortiz - 2010 projected: .267/.368/.492, 19.3 VORP; 2009 actual: .238/.332/.462, 14.5 VORP. Let me get this straight, 35 year old Alex Rodriguez will LOSE VALUE from his approximately 50 VORP performance in a partial 2009 to a 47 VORP across ALL OF 2010, but 34 year old David Ortiz is due to have a comeback?
On the pitching side, PECOTA's projections are slightly more conservative than many actual 2009 performances, but all of the conventional stat projections are satisfactory. For example, Beckett and Lester project to have ERAs of 3.60 and 3.89 respectively. Fans may disagree, but those are reasonable estimates and not substantively different from their 2009 actual ERAs of 3.86 and 3.41. Beckett projects to improve, in fact. PECOTA projects all 3 of the Yankees' big '09 starters to get worse in '10.
PECOTA does have a harsh projection for Papelbon, but that just proves not even computers can stand that jerk. Just kidding. PECOTA seems to be systemically poor at projecting relievers in general. Here are three examples, all from the projections:
Jonathan Papelbon - 55 IP, 2.84 ERA
Mariano Rivera - 55 IP, 3.31 ERA
Phil Hughes - 59 IP, 4.02 ERA
BP sees each of these guys losing AT LEAST one run per 9 innings from their 2009 performances. Ain't happening. The fact that virtually every reliever on either team, even Boone Logan, is projected to pitch no less than 40 and no more than 59 innings doesn't add credibility to the bullpen projections either.
While I wouldn't be shocked to learn that Baseball Prospectus is headquartered in Braintree, MA, I do not believe they are intentionally trying to make a downward projection for the Yankees. Rather, PECOTA consistently makes conservative estimates, and the Yankees seem to be penalized for having old players who are expected to contribute greatly to the team. A quick glance at the Red Sox confirms the absence of such a penalty, at least for one other team.
Guess VORP is as VORP does, huh? Of all the new advanced stats from the last few years, I like UZR the best and find it to be simple to understand; not that this helps the situation we're in with PECOTA. When comparing the Yankees with the Red Sox, I take a look at the rosters and just make a judgment call: both teams are pretty even in starting rotation and defense, but I give the yankees an edge in their bullpen and their offense. i think the Yankees offense will have a very minor decline from last year but will have improved pitching and defense, so i figure them to win the same number of games this year as they did last year. As for the Bo Sox, I think their offense will decline at a slighter greater rate than the Yankees will, their starting pitching is improved from last year and their bullpen will probably be about the same and their defense will be improved. I see the Yankees finishing first and the Red Sox grabbing the wild card. The Rays are good, but the Red Sox are better, and Theo knows how to build a team for his home ballpark.
ReplyDeleteAlthough I think Mike's complaints are generally reasonable, he might have tried comparing the 2010 prediction with a longer period than one year. Many Yankees did unusually well last year, so a prediction based on several years will naturally show a decline from 2009. OTOH Ortiz had an unusually bad year in 2009, so the same principle applies in reverse.
ReplyDeleteI have to confess that I'm not a big fan of UZR either. My concern with UZR is that it is not always meant to be used on a year-by-year basis. Instead, comments on my own posts have advised me to use the sum total of a player's UZR over his entire career. Due to the volatility of year-to-year performances for all but the best players, I'm skeptical of a statistic that explicitly can't be used on an annual basis. But I do recognize the importance of improved defensive metrics.
ReplyDeleteI agree with your overall assessment of the Sox and Yankees. When numbers and facts inform a decision I strongly believe it is ok to say that a prediction "feels right". My largest criticism of the BP predicts is that they don't feel right.
With regard to comparing the 2010 predictions to 2009, the argument also holds for many of the players entire careers, at least on the Yankee side. In 2009 Jorge, Jeter and Mo each had seasons that were better than their career averages. BP projects that each one will post one of his worst seasons ever in 2010, so the prediction is well below their multi-year averages as well.
Of the projections that give me pause, only A-Rod is close to his career averages, and even he comes in below.
The irony is that had their current projection been their original one no one would have complained. While the numbers are low I can't imagine so many posts across the web would have been dedicated to the predictions if the Yankees had been in first from the get go.
Man me i am just worried about Alex Rodriguez and his latest involvements with the authorities. Also, i don't think Matsui's loss will affect the yankees at all in my opinion.
ReplyDelete