that's what I'm trying to figure out here. if 20 hypothetical players are worse, I would expect around 20 hypothetical players to be better and the other 160 hypothetical players to be pretty damn close to where they always are If 20 are worse, why even assume 20 are better? That makes absolutely no sense. If a certain subset of players is performing worse, it is a mathematical certainty that someone is performing better. And yes, collectively, the 180 players would be performing better. If 20 players performance worse in clutch situations than they normally do in regular situations, how in the hell is there any certainty that 20 others perform better in the clutch than they do in normal situations? You must be confusing with me someone else. I've never argued that. If 20 players perform worst in high-leverage spots than their expected performane generally, the rest of the players collectively (All players minus the 20 "anti-clutch" players) must perform better in hig-leverage spots than their performance generally. Obviously, within that 180 players there is going to be a inevitable distribution of performance, some better, some worse, but the total collective performance will be better.