clock menu more-arrow no yes

Filed under:

Defying PECOTA: How the Colorado Rockies might beat their 2008 projections

New, 6 comments

Quickly, what's the most important position (non-pitcher) in the Rockies bid to repeat as NL champs this season? Second base? Catcher? According to Baseball Prospectus' PECOTA, our supposed 2008 starters at those two positions project to lose just over two games total from how the primary starters performed at them in 2007. Sure, two games are important, and I'd love to have us reverse that projection, but that isn't nearly the biggest loss BP is saying we take on the field this year and by this measure, they don't qualify as most important. Hawpe, Atkins, Taveras? Nope, those three have a combined drop of a half game in their projections from 2007.

Try our stars. Helton, Tulo and Holliday -who in 2007 combined for a 26.2 WARP1 according to BP- are projected to be just under sixteen wins above their definition of replacement level this year. According to Baseball Prospectus, we're expected to lose over ten wins at first, short and left field by trotting out the same players that we did in 2007. Holliday is expected to decline the most, dropping over four wins from what he gave us in 2007. Helton's projected to give up just under four and Tulo nearly two and a half. Don't blame the numbers or those that use them for this, the age curves and regressions that they are based on have proven very accurate in anticipating future performance. PECOTA also culls historical comparisons from a database of thousands of players to improve its accuracy. So it seems that one of the biggest issues for the Rockies in 2008 that nobody's talking about will be whether Holliday, Helton and Tulo can continue to be all-league caliber rather than just the very good players the numbers seem to be saying they are.

One thing to keep in mind (as Heltonfan points out here) that because of this profiling for similar players, PECOTA seems to have developed a bias against Holliday -whom it's fairly certain is a reincarnation of Ivan Calderon or even more far-fetched, Marty Cordova- that other projections aren't seeing, but all projections I've seen seem to have the team's biggest drops in value coming from the heart and soul of our lineup rather than the peripherals.

So what I'm saying is that, first of all, we shouldn't be surprised when the Rockies show up under-performing according to our expectations in season simulations, but also that it might be interesting to explore the issue of why and how these projections might be undercutting us. The question is that whether or not Rockies fans have any more reason to believe that their team will outperform the projections over any other team's followers. After all, it's probably true of at least twenty-four fan bases (those not expected to come in first in their division) that they will find some fault with whatever set of projections are out there, but do Rockies fans have particular reason to be confident that we will come in ahead of the curve? Obviously, PECOTA's Holliday projection shows an example where this might be the case, as Holliday's no Calderon, and he's certainly no Cordova, yet for the second year in a row, Marty's shown up as one of Matt's four most comparable players according to the projection system. Why the disrespect, then, and could the comparisons be shedding clues to other places the Rockies might be getting underestimated?

The disrespect in comparisons seems to stem from the complications that arise from adjusting for Coors Field, which usually -as seems to be the case for PECOTA and many other projection systems- seems to amount to taking the Rockies home field out of the equation altogether. Marty Cordova's career OPS is .778. Matt Holliday's career OPS outside of Coors is also .778. Coincidence? I doubt it. This fix of heavily weighing Rockies hitters performance while on the road, while simple, becomes problematic in certain cases and Holliday's an excellent example of that. As a team, Rockies hitters put up a .261/.336/.395 line away from Coors Field last season. Holliday put up a respectable .301/.374/.485. When at home, the team (including Holliday) obviously got better, hitting .292/.372/.480, around a 15% increase in OPS from their road stats. Holliday, however, hit a whopping .376/.435/.722 at Coors, representing an increase of over 33% from his road OPS. For his career, Holliday's improvement at home shows a dramatically large spike compared to what the team or Rockies opponents achieve when visiting Coors. The result will have projections that don't take the variations in Coors Field production by different types of players into account drastically underestimate Holliday's output in half of our games.

PECOTA seems to have fallen as heavily into this flub as any projection system I've seen. The end result is that they project our lineup to be equal in value on the season to Arizona's, which is a laughable premise. Are there other players who might be similarly overlooked? Almost to as much of an extent as with Holliday, Troy Tulowitzki appears to be falling into the same category. One pretty good clue is the dramatically high BABIP (Batting Average of Balls In Play) at Coors these two show, Holliday has a career BABIP of .386 at Coors and Tulo already hits .370 when he puts the ball in play while at home. The significance? BABIP's of .370 or more at Coors Field are more reminiscent of Larry Walker (.375 at Coors for his career) than Vinny Castilla (.330) and a solid indication of the legitimacy of a player's bat. Using Coors BABIP as a guide, the third member of the important trifecta I talked about initially, Todd Helton could also be getting the shaft from projections that aren't looking closely enough.

Significant Rockies BABIP's at Coors:

Matt Holliday .386
Andres Gallaraga .378
Larry Walker .375
Troy Tulowitzki .370
Todd Helton .362
Ellis Burks .362
Eric Young .356
Dante Bichette .352
Garrett Atkins .348
Neifi Perez .333
Brad Hawpe .333
Vinny Castilla .330
Yorvit Torrealba .311
Clint Barmes .296

As a side note in researching this post, one effect of the humidor seems to be that it's spread the ranges of BABIP's at Coors, as the team's better hitters put up just about as gaudy stats as they were doing in the pre-humidor days, but lesser hitters aren't getting quite the same benefit anymore.

Alright, I'm just going to put this food for thought up for digestion right now, and I'll probably revisit it later. While what I'm seeing with this makes sense to me, since I'm admittedly not the most mathematically inclined, maybe some of you who are can tell me if I'm barking up the wrong tree, here.