Your example of a team going 10-20 and then winning the league with 95 wins is disturbing if that is common.

Have you tried to use the stats from 30-40 launch leagues to identify possible problems with your SIM? You could statistically analyze the actual vs GAME stats for each player using each LL as a rep - I'm not sure if an ANOVA procedure or a chi square test would be more appropriate, but either would probably work.

For players who performed differently from expectations (statistically significantly different), you could check for sections of your code that may be responsible.

I think perhaps you are trying to take too much into consideration resulting in everyone falling towards the overall mean. By adding variables for parks, rh vs lh, night game vs day game, wind blowing in or out at Wrigley, etc. you could be causing a lot of random "noise" that overshadows the actual player stats.

I realize you don't put in wind direction and some of that, those are just examples. You do seem to be trying to account for a lot of variables though besides the basic stats. I don't know what all you put in your SIM (maybe you do randomly select a wind direction and velocity for each game?). My point is that by putting in too many totally random variables, you create chaos and unpredicability.


Have you analyzed statistically the accuracy of your SIM?


Don't get me wrong, I really like the game, the concept, etc. Just making some suggestions.

Dave