THE BOOK--Playing The Percentages In Baseball

Great read and good discussion.  I am most interstd in the park adjustments/factors section, which was Part 2 I believe.

Obviously multiplicative park factors won’t work, but if you put players into four categories (GB hitters, FB hitters, LHB, RHB) can you still then apply park factors as a whole to all players of each of those groups?

For example, let’s say player A is a RH GB hitter who has 50 total walks and the walk PF for was .90, you would take 25 (half the total) and divide by .95, which is 27.8 and add that to the other half, which is 25, for an adjusted total of 52.8.  Would you apply that same logic to every member of the team that is a RH GB hitter?  Because techincally that would be using the park factor for everyone (which is multiplicative and wrong).  If not, I don’t know how you can even use park factors/adjustments (at least now).  I assume that separating the hitters into these categories minimizes some of the problems.  Maybe if someone can provide a specific example of how you would adjust a player’s statistics when moving to a new team (Lowell from FLA to BOS - interesting case b/c Lowell is a doubles, pull hitter and Fenway is a doubles haven to left).

------------------------
Oh, and in order to determine a player’s park adjusted statistics when moving to a new team, don’t you need to do these three things: 1) Gather his actual statistics, 2) divide those statistics by the appropriate park factors of the team he left, and 3) multiply the new statistics by the park factors of his new team.  Walks and SO would be calculated per PA and hits, doubles, triples, and homeruns per BIP.

I could see problems with that myself.  In 2005, it looks to me, Richie Sexson’s power was affected less by Safeco field’s Left Field than other RHB.  He had a pretty good year, and if you look at his hit charts that year he was hitting the ball to all fields consistently. 

I’m almost certain this was the case for Bret Boone when he was with the Mariners.  I remember looking at his hit charts (MLB.com doesn’t have his anymore), and he hit a lot of his home-runs out of right field, which, means Safeco may not have affected his power much at all.

Oh, and I have no clue how year to year hit distribution works on average, I noticed Richie Sexson is pulling the ball a lot more this year. Alex Rodriguez is hitting the ball out of left the ball more too, but in past years he pretty consistently hit it out of all fields in whichever park he played in. 

Has anyone ever done a study on year to year hit distribution per player?  I have no idea.  I suspect it’s folly to try and apply that to anything.

Yeah, that may have to be considered too, although where they hit the ball in one year may have a random variation as well.  I’m not sure, though...hopefully others can add more.

What do you guys think?

I would imagine, another problem might be the reliability of L/R park factors. Due to sample size, particularly with Left Hand splits, I’m not sure you can get a large enough sample size, even w/ 3 year data, to reliably factor some components.

The year-to-year hit distribution is applying the granular data that the group was talking about in Part 1...and sounds like a good way of park adjusting for each individual player.  I’m just not sure how reliable a player’s hit distribution is over a couple of years or if it’s reliable at all.  I just don’t know.

MGL, is that the way you park adjust for players, though?  Do you use the granular data?  Or, for example, do you apply a walk park factor (like in my example) to everyone on one team (which is multiplicative, right?)?

Yeah I’d be interested in someone letting us know the reliability of breakign it down into LHB/RHB and GB/FB and how many years it takes for that park to be considered to have reliable results.

Along with these park adjustments, don’t you now have to penalize NL hitters when they move from the NL to the AL b/c mgl clearly showed that the difference is at least 10 games per season between leagues...This may be taken care of when you make league adjustments I guess…

I guess there’s no way yet to edit your post, so I’m sorry for the three consecutive posts…

Also, as mgl stated in the roundtable, park adjustments only apply to those players who change teams correct?  But then mgl says that if a player has played in different home parks in your historical database, park adjustments are necessary.  So does that mean that Scott Rolen’s projection for 2006 needs park adjustments?  He’s played for St. Louis for awhile now, but that’s not the only team that would be included in his historical stats (Phillies, etc.).

MGL I was wondering if you could explain some of this to us.