Andrew got me on the regression equation kick. Here's one for predicting "conference wins"
Conference Wins= 18.44 - (.141)x1 + (.654)x2+ (1.786) x3
Where x1= Avg skills of team (Total Avg minus durability, work ethic, rebounding)
x2= Avg rebounding
x3= Avg Offensive knowledge (where 12= A+ and 1= D)
It's not pretty b/c the avg skill is negatively sloped (making y intercept 18.44) but it adds significantly to R^2 which is .89. I also created a very loose category for offensive knowledge that accounted for the fact that it is much harder to move players though the "A" category than C and D category. I therefore made D="1" C="3" B="6" and A="10." I averaged in only those players who played at least 10 minutes a game to create an average for the team. This turned out to be a very important factor b/c it was not correlated to the rebounding stat- where nearly every other stat I tried to put in was correlated with Rebounding.