Posted by katzphang88 on 5/3/2013 5:27:00 PM (view original):
Posted by gt_deuce on 5/3/2013 4:36:00 PM (view original):
I just thought of another problem with all the methods of "normalization" I tried (and/or those suggested in this thread):
If you normalize as fatmoose suggested above - a methodology I also tried, but had forgotten about - you're saying that the best QB is always as good as the best RB is always as good as the best WR and etc. This is not true. Additionally, because of the big discrepancy in STDDEV at some positions, you still get a very dissatisfying mix of "Overall Top 50/100" recruits.
I understand what you are saying, but in the way the GUESS ratings are set now is that the best QB can never be numerically better than the best RB or WR because WIS just doesn't give them high enough values. Perhaps in the scheme of the game - the best players out there to recruit are OL or DL (look at the NFL draft) - but in GUESS reports they will never be #1 either. Bob is trying to counter the differences in WIS ratings to make the GUESS ratings more of an indication of the best recruiting class - to make his game within the game more enjoyable to the coaches (Thanks for the effort Bob - I much appreciate it.)
I personally don't look at the player scores very closely. I look at the rankings (overall and in class) to see how they stack up. When version 3.0 rolls out, with all the various player roles, it will be harder to compare values across all positions. The best blocking TE in the WCAA may score lower than the best receiving TE. Do you penalize the team with the best blocking TE? I say give them both XX points for having the #1 and XX-1 for #2 and so on. The values for each season will stay the same, QB, P, K and blocking TE's won't have to be normalized and we can still enjoy Bob's comparisons for bragging rights.
A little sneak peak about GUESS REPORTS 3.0 or whatever - I'll be including ratings for the position roles too. Team and class scores are going to be based on the base position, I expect, otherwise it's going to take six hours for the scoring algorithm to run.
The idea of scoring the recruiting class based on the RANK of the recruit, rather than their rating, is interesting. I can think of a couple of different ways to implement it. I might have to play around with that some this weekend.