I like APM for measuring defensive impact as well.
]]>For evaluating individual defense, you'd have to combine looking at Def. Win Shares with +/- because its hard to precisely determine how much of a win a guy like Tim Duncan contributed vs. Bruce Bowen in the Spurs defensive heyday.
Allin all, its hard to 100% go to the bank and earn interest on stats except for baseball. I'll take a player's WAR rating - wins above replacement - and earn a good return on that one in a bad economy(well ok maybe that's taking it too far haha). I certainly favor Def. Win Shares in basketball over Off. Win Shares since as you alluded to it seems off win shares favor big men who grab off rebounds and guys who go 3 for 5 on tip ins like Jeff Foster vs. Jermaine O'Neal...or even Tyson Chandler who I think is a more valuable defensive player than offensive.
]]>You seem like a very reasonable person overall, thank you for contributing to this discussion. I think some of the things you said are very interesting.
In general I think it is best to use a wide assortment of metrics to rank players, because some stats can be goofy at times. If you want to use PER don't forget to adjust for usage rate (especially if you're comparing players on the same tier). For instance Jordan has 27 PER some seasons, but on a massive usage rate. He has higher peaks than Kobe but not all of Jordan's "27 PER" playoff seasons are superior to Kobe's "25-27 PER" playoff seasons (2001 and 2009). That's just one example. I like basketball on paper SPM because it incorporates usage, defense, offensive/defensive rating, minutes per game, +/- fundamentals.
#26 I agree that Jermaine is a better offensive player, there is definitely no end-all stat. Some stats weigh the usage-efficiency tradeoff differently, in this case OWS penalizes Jermaine heavily for his 100 offensive rating.
My ideas would be to use Neil's ranking system, but adjusting for total minutes, era (this is particularly difficult), defenses faced as well (RAPM addresses that to a degree, perhaps we need more metrics with that feature).
Certainly you're correct, comparing All-Stars to role players is inappropriate at times.
]]>This is a spot-on list. Great work.
]]>