Reading Paul’s link to Neil Payne’s article ranking point guards by Alternate Win Score motivated me to see what the hell Alternate Win Score might be. Turns out it’s the same thing as Dave Berri’s linear weighted metric Win Score — (Win Score = PTS + STL + ORB + 0.5*DRB + 0.5*AST + 0.5*BLK – TOV – FGA – 0.5*FTA – 0.5*PF) with additional weights regulating the relative value of offensive and defensive rebounds as well as field goal attempts and free throw attempts. (Oh snap! I’ve got NBAtv on, and Brandon Bass just earned a Tommy Point! Hell yeah!) Basically some smart math guy discovered that making these adjustments delivered a more predictive result and eliminated the need to correct for positions.
Here’s a quick spreadsheet I put together using career per-game data beginning from the 3 point era – initially sorted by Win Shares and then resorted by AWS after I performed the calculations.
For those too lazy to click the links, here’s the breakdown by position with overall ranking in parentheses.
PG – Magic Johnson (3), Chris Paul (7), John Stockton (25), Kevin Johnson (32), Isiah Thomas (43)
SG – Michael Jordan (1), Clyde Drexler (15), Dwyane Wade (16), Kobe Bryant (23), Vince Carter (34)
SF – Larry Bird (2), LeBron James (4), Julius Erving (12), Alex English (18), Shawn Marion (20)
PF – Charles Barkley (5), Karl Malone (10), Kevin Garnett (11), Dirk Nowitzki (13), Chris Webber (19)
C – Hakeem Olajuwon (6), David Robinson (8), Shaquille O’Neal (9), Tim Duncan (14), Kareem A-J (17)
Boy, we’ve got a lot of the right names in there, don’t we? The order is funky, but that’s because the stats aren’t pace-adjusted or minute adjusted. Still if you consider that AWS is an offshoot of a metric that once measured Dennis Rodman as a more valuable member of the Bulls than Michael Jordan, it’s a nice step. It certainly isn’t meant to be used to measure players between generations the way we have here, but I’m actually pretty fond of the results though I had nothing to do with them. I’d probably get better results working year to year or in 5 year clips instead of as an all-time ranking this way.
I can’t figure out how Clyde could possibly outrank Wade, but I think it might have something to do with offensive rebounding and low turnovers – see Paul’s article regarding the benefits of Drexler’s possession creation here. Kobe of course is almost always rated lower than we’d expect by metrics. Too many misses and turnovers, plus low-production and low-efficiency in his teen years. If I set age limits, he would move up. So would LeBron and Garnett probably. Though if I set maximum age limits some of that might be reversed. Actually maximum age is probably why Pippen is below Marion.
No one position is massively underrepresented at the top. No decade is missing (though the newer players like Rose and Durant didn’t have the opportunity to make the list since I made it based on career Win Shares.
One of the things I like about the Win Score model is that it’s accessible to the mathematically challenged among us and has a very comprehensible logic about it. There have always been a couple of fundamental flaws with the formula though. One problem is the way the stat considered a shot to be as bad as a turnover which the Alternate formula corrects by factoring the 30% chance of an offensive rebounding. The other issue is that the stat weighted rebounds with no regard for the diminishing value of grabbing a defensive board. The Alternate formula factors in the likelihood of a team losing a rebound on either side of the ball and adjusts by that percentage.
I’ll work with this model a little more in the future to see if we can factor AWS into the Grand Ole Double Dribble NBA Player Ranking System (TM).