Playing with PECOTA
Baseball Prospectus has released their PECOTA player cards, and a half hour of clicking through all of the Detroit Tigers there are definitely some interesting highlights. But before we get into the neat stuff, a little bit about PECOTA. PECOTA is a projection system based on finding comparable players. Similarity is calculated based on performance, usage (playing time), and body similarities (handedness, body type). The system then uses the pool of comparable players (which could be as many as 100) to project what the player in question will do. One of the data points available is the similarity index. The higher the score, the better the pool of players matches the player in question. A score over 50 is pretty good, and less than 20 is pretty bad.
Now PECOTA is for BP subscribers only, so I won’t be laying out all the stats provided. I will just be hitting some highlights. (if you want to see what the cards look like, the White Sox are offered for free.
PECOTA is pretty high on Bonderman, and a quick glance at his comparables shows why. Some of the pitchers on his list include Joe Coleman, John Smoltz, Jake Peavy, Tom Seaver, and Catfish Hunter. His improve rate for next year is 77%, with no discernible decline until 2010.
Last year was enough of an anomaly that PECOTA doesn’t know what to make of Jones. His similarity index is only 35. That said, the model thinks he’ll be a couple worth a couple wins above replacement level.
There aren’t very many pitchers who are still pitching in their 40′s. That’s why Rogers similarity index is only 21. The advantage is that to be pitching in your 40′s, you must be doing something right. So his comps are Warren Spahn, Jamie Moyer, Tom Seaver. Not surprisingly, the model doesn’t look favorably on Rogers. His median projection is for an ERA north of 5 and essentially replacement level production.
Justin Verlander & Joel Zumaya
The model likes both of these guys. The median projection for Verlander has his ERA at 3.98. In Zumaya’s case, it is 4.51 with 8.7K/9. Either season would probably be good enough for rookie of the year.
That’s it for the pitchers. I’ll take a look at the hitters next.