how did what was a constructive thread turn into 20 pages of ****?
at the risk of repeating what was said, i think we all agree that in a single game, overall rating is a poor indicator of who will win. a 50 points overall rating difference can easily go the other way, or given equal overall ratings, one team can be in the ballpark of a 98% favorite.
but, in a long term sense, overall rating is a somewhat reasonable measure. on average, the 620 rated d2 team is better than the 600 rated d2 team. despite the countless counter examples, it is not a useless measure. consider the OR ratings for d1 teams. they are based purely off per-player overall rating and class, and have proven a fairly accurate indicator of the quality of a team.
using overall ratings in rankings before the season started is fairly reasonable to me. its the easy way out, the cop out, but if that component is going to 0 by season's end, its largely irrelevant. using overall ratings to me is step 1. step 2 is throw out the highly useless in-game ratings; dur, work ethic, lp/sb for guards, per (maybe) and bh for bigs. step 3 is to actually rank players in an intelligent manner. i used to use a recruiting program to rank players to save me time, and i came up with fairly complex formulas to rank each player, that ranked them within a small margin of how well i could rank them, in most cases. this is where i think WIS should be - step 3, actually assign the players a score based on their quality. but, if they are busy, i think step 2 would be acceptable. step 1 is just too ****** and its too easy to get to step 2.
to me, step 4 is ranking players more than 1 way. for example, how do you rank a shooting point guard vs a non shooting point guard? the best equations for one are less valid for the other. what about at the small forward position? its a tough problem. you have to guess who will play where, assume some level of coaching prowess to put the right guy in the right place, and rank accordingly. i think this is too hard for WIS to undertake. also, i don't think simply being the programmer of the game is enough to do it competently. this is not a knock on seble, but as a guy with internal code knowledge, if he couldn't come out and beat the likes of OR and lostmyth, he probably lacks the working knowledge to undertake step 4. so, step 4 is the last place i think WIS should be, on the rankings.
step 5 is incorporating synergy between players and the system at hand. this is to me essentially as complex as writing the sim engine. however, if you write a decent coach to set appropriate ratings for a team, and simulate them all against each other, you can approximate doing step 5 the real way. a clever programmer could do this way faster than step 4, so, to me that is the ultimate ranking system. writing a decent coach is easier than it sounds, it can be done optimally without understanding the sim engine in the slightest. you just have to cleverly pit pseduo random strategies against each other until some emerge as better, and then further randomize off of that and repeat.