The 06 Kobe comparison is extremely apt: both players are putting up staggering counting stats (with comically high usage rates) on fair to middling teams, and neither player is very well regarded by composite stats. For example, Kobe Bryant produced fewer Win Shares than (among others) Chauncey Billups did for the 64 win Pistons in 2006. The gut reaction is that that’s wrong, Win Shares are wrong, the whole darn system is WRONG… but guts aren’t brains. We can evaluate Win Shares across decades and see how well it correlates with actual wins, and the answer is quite well. If the only time the methodology appears not to work is for players we like, maybe the problem is that we’re letting the like get in the way, no?