Sunday, July 09, 2006

How is the projection system doing?

I could go over the hits and misses for all players, like the Stats and Bill James handbooks have always done when looking back at prior year projections. I could pat myself on the back for predicting Andruw Jones to have an Andruw Jones type of season, and wonder how I missed David Ross turning into Mike Piazza. But I won't. If I want to learn something here, and make better projections in the future, I think its best to see how the workings of the system treat groups of players.

Here's the summary in spreadsheet form: Projection player groups

Lessons learned:

1. For the youngest players, age 21-24, there is a fairly substantial HR boost applied to these players. So far at least, they aren't quite hitting it. I may have to tone it down a bit.

2. For older players the age adjustments may be increasing strikeouts a bit too much.

3. The projections for speedsters look right on the money.

4. The players projected to hit for the most power are hitting for more power than projected, the weakest players are hitting for less power than projected. The same thing is happening for the players who take the least and most walks. What this means is that I'm regressing walks and homers too much to the mean. If I tone down the regression and allow the player's actual HR and walk rates to have more impact, I should get better results.

0 Comments:

Post a Comment

<< Home