You are currently viewing archive for May 2010
Posted By Sean S

Earlier I said that I would not have updated pitcher projections ready this season.  I lied.  I was able figure out how to do what I needed to get them done, and now you can see them too.  Hitter projections are starting to look different, substantially so in some cases, from preseason projections.  In preseason I was using years 2009-08-07-06 for hitters, with 2009 weighted 100% and the others proportional to that.  After updating for April, what I did was still weight 2009 as 1.00, keep all the other years, and effectively treat the first 30 games of 2010 as an extension of the 2009 season.

Now that I've got nearly a third of the new season underway, I have lowered the weight of 2009 and previous years, and dropped 2006 altogether - CHONE has typically used 4 seasons of data.  CHONE was somewhat skeptical of Jason Heyward's ability to contribute right off the bat, but now is a big believer, he's one of the best in the game at only age 20.

For pitchers, the guy I wanted to see is Strasburg.  At a 4.37 ERA, CHONE is obviously hedging quite a bit.  Keep in mind that it knows NOTHING about his college career, and nothing about his velocity ( though I'm working on that part).  It sees a great 50 innings in the minors, but doesn't know if he's really that good or just lucky.

Sorry for the lack of posts and responses to questions.  One guy asked about whether he should only use preseason projections to update team odds if pitchers were not available.  The answer is, have at the pitchers.

Posted By Sean S

I've done them, though it was quite a chore, more than I expected.  It takes enough effort to get the data from the MLBAM gameday files into a form that I can use for projections, plus time to try and make sure I've got players with their new teams.  I doubt I'll be able to do too much with this, I'm going to shoot for monthly updates.  I may not get to the pitchers at all this year.


There was a methodology change, I was doing the regression before I applied the player's current age to the projections, and this resulted in lower than they should be projections for very old and very young players.  The way I think it should be done is all players are regressed to their league's average, regardless of age. While the entire population of 42 year olds should hit worse than the population of 27 year olds, the 42 year olds who are in the major leagues should hit as well as the 27 year olds in MLB.  This has generally been true over the course of history.


The above is part of the reason Jason Heyward had a -6 projection in my preseason file, and now is at +13.  He's the perfect storm though, what he also has going for him is being regressed to the major league mean instead of the AA mean, and finally, he's done a pretty good job hitting for the Braves.


A few other guys off to great starts and upping their projections are Paul Konerko (+11 to +16) and Kelly Johnson (+2 to +8).  Most players probably haven't changed much though.  Want to know who else has a significant change?  Then look through the file, and have fun.




User Profile
Sean S
Maryland, USA


You have 383387 hits.

Latest Comments