I get Stathead daily emails from Sports-Reference, and something has been puzzling me about the NBA this season, and I'm wondering if you have the time to make a post to explain this to me (and anyone else who might be interested).
Chet Holmgren is having the strongest season of any rookie this year. (8.2 Win Shares - leads all rookies, Derek Lively is second with 4.6!) He's ALSO leading all rookies at 3.2 VORP. That, again , makes sense to me. Usually, ranking highly in one of those statistics means ranking highly in another.
But there's a puzzle. Because the second-leading VORP among rookies belongs to Victor Wembanyama, at 3.0 VORP. Now, when I look at that number, and I see how much Wemby has played this season (63 games, 29 minutes per game; compares favorably to Holmgren's 70 games and 30.1 minutes per), I'm left to wonder: what on earth is that Win Shares is or isn't capturing in Wemby's performance that has him at a relatively miserable 2.8 Win Shares? His Box Plus-Minus (4.4) is actually better than Holmgren's (4.0), and while he hasn't played AS many minutes, it's not like he's sitting out half the games. So what gives?
I'm guessing it's because Chet's Offensive Rating is 124 and Wemby's is 104. (League average is ~116.) On offense, Win Shares is based on outperforming 92% of the league average ORtg:
One of the longstanding criticisms of WS is that it doesn't account for any tradeoff between Usage and efficiency, so Wemby's 31.8% Usage Rate doesn't really get him anything extra aside from a higher multiplier to his per-possession ORtg gap vs 92% of average (which actually hurts because he is below that threshold). Because of this, high-usage, low efficiency players like Wemby are always undervalued by WS and low-usage, high efficiency ones like Lively are overvalued.
eRAPTOR has a bit of this problem, too, as it views Wemby as a -2.3 on offense (BPM has him as a +1.5). Although with eRAPTOR I think it's actually more inferring that Wemby must not be having all that great of an impact on the offense when the Spurs' ORtg is actually higher when he sits:
eRAPTOR is looking at him and saying "hmm, the league's 27th-best offense actually gets better when this guy sits, and his individual efficiency is way below average... are we SURE he is a good offensive player?"
Is that right? I think something like BPM will end up being predictive of Wemby's future value. But it's always tough to measure the present performance of players who are clearly good enough to get their own shot whenever they want at a young age, but have subpar efficiency and turn the ball over a lot. Are they helping or hurting? Truth is, it doesn't really matter; the 2024 Spurs were never about the present day anyway.
Thanks for the explanation. I wasn't aware of the 92% threshold. Because, as you note, BPM thinks Wemby is an inefficient offensive player, too. But, you know... you can see that he's still more than worth it, if you look at how great his defense is.
I don't know that I've been aware of a case, prior to this one, where these advanced NBA stats were THIS divergent - Wemby might be the BEST rookie, or he might not be in the top 10. That's pretty strange. So thanks for the quick explanation.
As for my question, I'm not REALLY thinking about future productivity; more how I should view the ROY race this season. It's seemed to me all season to be Holmgren's to lose, but obviously Wemby has been a bigger story, and so looking at his metrics being so all over the place made me curious. Thanks again for the explanation, which was both quick AND thorough!
Any chance you are publishing a 2024 march madness forecast? I relied on the 538 bracket forecast for over a decade. My 538-based brackets were usually in the top 85th+ percentile nationally, and I won several local contests.
Neil:
I get Stathead daily emails from Sports-Reference, and something has been puzzling me about the NBA this season, and I'm wondering if you have the time to make a post to explain this to me (and anyone else who might be interested).
Chet Holmgren is having the strongest season of any rookie this year. (8.2 Win Shares - leads all rookies, Derek Lively is second with 4.6!) He's ALSO leading all rookies at 3.2 VORP. That, again , makes sense to me. Usually, ranking highly in one of those statistics means ranking highly in another.
But there's a puzzle. Because the second-leading VORP among rookies belongs to Victor Wembanyama, at 3.0 VORP. Now, when I look at that number, and I see how much Wemby has played this season (63 games, 29 minutes per game; compares favorably to Holmgren's 70 games and 30.1 minutes per), I'm left to wonder: what on earth is that Win Shares is or isn't capturing in Wemby's performance that has him at a relatively miserable 2.8 Win Shares? His Box Plus-Minus (4.4) is actually better than Holmgren's (4.0), and while he hasn't played AS many minutes, it's not like he's sitting out half the games. So what gives?
I'm guessing it's because Chet's Offensive Rating is 124 and Wemby's is 104. (League average is ~116.) On offense, Win Shares is based on outperforming 92% of the league average ORtg:
https://www.basketball-reference.com/about/ws.html
Because Wemby is below that threshold, he actually has negative offensive WS this season:
https://www.basketball-reference.com/players/w/wembavi01.html#advanced
One of the longstanding criticisms of WS is that it doesn't account for any tradeoff between Usage and efficiency, so Wemby's 31.8% Usage Rate doesn't really get him anything extra aside from a higher multiplier to his per-possession ORtg gap vs 92% of average (which actually hurts because he is below that threshold). Because of this, high-usage, low efficiency players like Wemby are always undervalued by WS and low-usage, high efficiency ones like Lively are overvalued.
eRAPTOR has a bit of this problem, too, as it views Wemby as a -2.3 on offense (BPM has him as a +1.5). Although with eRAPTOR I think it's actually more inferring that Wemby must not be having all that great of an impact on the offense when the Spurs' ORtg is actually higher when he sits:
https://www.basketball-reference.com/players/w/wembavi01/on-off/2024
eRAPTOR is looking at him and saying "hmm, the league's 27th-best offense actually gets better when this guy sits, and his individual efficiency is way below average... are we SURE he is a good offensive player?"
Is that right? I think something like BPM will end up being predictive of Wemby's future value. But it's always tough to measure the present performance of players who are clearly good enough to get their own shot whenever they want at a young age, but have subpar efficiency and turn the ball over a lot. Are they helping or hurting? Truth is, it doesn't really matter; the 2024 Spurs were never about the present day anyway.
Thanks for the explanation. I wasn't aware of the 92% threshold. Because, as you note, BPM thinks Wemby is an inefficient offensive player, too. But, you know... you can see that he's still more than worth it, if you look at how great his defense is.
I don't know that I've been aware of a case, prior to this one, where these advanced NBA stats were THIS divergent - Wemby might be the BEST rookie, or he might not be in the top 10. That's pretty strange. So thanks for the quick explanation.
As for my question, I'm not REALLY thinking about future productivity; more how I should view the ROY race this season. It's seemed to me all season to be Holmgren's to lose, but obviously Wemby has been a bigger story, and so looking at his metrics being so all over the place made me curious. Thanks again for the explanation, which was both quick AND thorough!
Any chance you are publishing a 2024 march madness forecast? I relied on the 538 bracket forecast for over a decade. My 538-based brackets were usually in the top 85th+ percentile nationally, and I won several local contests.
Unfortunately, I just don't have the bandwidth to resurrect something like the 538 model this year :(
As a replacement, I think I'll be using the ESPN BPI model: https://www.espn.com/mens-college-basketball/bpi/_/view/tournament