I'm not sure what you mean by factoring over the entire season, but if you mean what I think you mean, you have to look at the Stamina and Durability ratings together to determine how many plays they can normally be in and for how many games you can expect them to play in. Right now, the Stamina is determined on a per game basis, which means that we can not allow someone that had 30 rush attempts in 16 games run 30 times in 1 game if he sits the rest of the games. It means you can expect him to run about twice a game (i.e. not much use at all except as a 3rd or 4th back). A big reason for doing it this way is because we do not have seasonal fatigue in the game. When we do get it in, we can potentially allow you to look at Stamina a little differently.
The ratings DO drop as they go over their attempts. If players are overperforming it is because the formulas are not quite right, and this is something we will conitinuously work on as the game goes on as you just saw with the passing changes. Now, it does not start to drop EXACTLY when they have more carries or attempts than their Stamina dictates, but rather we look at the number of plays in the game and number of touches and calculate factors for their ratings. This is so your star player doesn't drop significantly at the end of the game just because he goes a little over his touches.
One thing I have pondered about doing is pushing the ratings more into the background and displaying "prorated normalized" stats, kind of like what SL Baseball is displaying. So instead of all the Stamina and Durability and other ratings, you might see a guy that played 12 games be expected to play 16 games, and instead of 20 rushes/GP, he might be more towards 23 rushes/GP. I think this might make it more clear and easier to compare players. The upside to ratings versus prorated stats is that it took the focus away from "hard" expectations. SimLeagues are based quite a bit on possible results and not necessarily on expected results. You expect Rudi Johnson to run for 140 yards against Buffalo, not 88, but it happens, and not necessarily because they have their Coach's Settings "right". Ratings were meant to be more of a comparison guide, like an 80 rated RB is generally better than a 70 rated RB. However, I think ratings have their own misleading assumptions, so I may as well do the stats thing.