I know Mike Bravard is proud of all the little statistical details PB puts into its ratings, but it seems to me that if you try to take too many things into account, you end up with just "noise" instead of really evaluating performance.

It seems to me like the game module also has this problem which causes players to have average performances much different from what you would expect from their stats. By trying to account for lefty/right, individual park stats, gb/fb for pitcher, gb/fb for batter et al.; what happens is that some of the statistical data just doesn't have enough volume of data to be meaningful.

For instance, this is why last year I was able to pitch Doug Nickle in a game where he went 17 innings and struck out 27 while in real life he only pitched about 5 innings. He wasn't as good as his "statistical" data would indicate but there just wasn't much data.

The same occurs with trying to use the data for pitchers in having runners steal and combining that with catchers data to make Pudge Rodriguez have trouble throwing out 40% of the runners against him last year when he had a 9.9 arm. Often a pitcher may only show 3 or 4 batters running against him so the results don't really reflect anything except random chance.

It also may enter into things like trying to combine the team gb/fb ratio to determine an outfielders range (although if it is team gb/fb and not individual pitchers that should have a fairly solid statistical basis?).

The lefty/righty stats are very often based on pretty limited at bats vs one or the other pitcher...if lefty righty stats are going to be used for a hitter, perhaps a 3 or 4 year average for the player would be a better way to evaluate how he hits vs lefties or righties. For players with limited data over the last 3-4 years (rookies & utility players), I'd say just use his overall average and adjust it for the "overall" lefty/righty stats for the league to get his lefty/righty breakdown. I think it would more accurately reflect reality.

Same thing for pitchers. Some righty relievers in PB look to have great stats vs lefty batters, and some lefty pitchers in PB look to have great stats vs righty batters. Reality often is they just have faced very few opposite handed batters and they are probably not nearly as effective as the very limited stats would make one believe. Just as for PLATOON hitters who have few at bats vs a type of pitcher, the pitcher who mostly faces one type of batter in real life shouldn't have PB stats that encourage you to use him against the other type of batter. For example, in this case, I'd suggest a league average for lefty pitchers vs righty batters and use that to adjust the pitching stats for the performance of limited use pitchers - perhaps taking into account how good they were overall and combining that with the league average to get performance vs the limited stat direction (lefty vs righty).


Over time (more data) players data tends to move towards the overall average, especially for players who are well over or well below the average but with small sample size data. Using league average data to help adjust for limited use areas will probably result in a SIM that more closely resembles reality.

IMHOP