BuzzerBeater Forums

I decided to introduce a randomized 'Aggression' number, for each player, value 1-5. totally random, but equal distribution.

Random Aggression value correlated with FP48 +0.16
well at least i got a positive correlation!

I gave my aggression statistic more of a bell-curve distribution (that is, more players have an 'average' aggression level, and less players have the extreme's) - I also increased this value distibution to match BB stat ranges 0 -20.
Doing this, i increased the correlation to about 0.25,

I don't see the point in this. obviously a totaly random value can't correlate with anything. You could have called that random skill fluffy candiness and still would have gotten the same result. You obtaining a non zero value is due to your small sample size.
But good job on the rest though. There really doesn't seem to be a big effect of any of the visible skills.

I took a guess with regards to the agressiveness, thinking that it would be a bell-curve distribution. Logically, I thought that BB wouldnt assign 50% of the players with a high aggressiveness, rather i guessed that Most players would have normal aggression, with the remainder of players either being less-aggressive, or more aggressive (hence bell curve)

And yes your right, randomly assigning it meant it could be called any kind of stat. Using the radom aggression assignment I was able to highlight that, even a ranomly asigned number has more correlation than the visible skills. Thus proving the argument a little bitmore, that being that OD and H have no correlation. ( I could have used anything here really, Height, hair colour, team uniform snazziness rating. )

Yer 50 players wasnt exactly ideal, but it was enough to give an indication of whether further data collection was warranted. Which, IMO, is not. :)
Thanks for the feedback.