RealGM 2017 Top 100 List #84 (Rasheed Wallace)

[Owly]

If you know PER a 15 PER is average. A 10-11 PER is a replacement level player. WS thinks say an 8 PER player should get some credit for wins. VORP thinks they are costing the team wins, again super simple.

A minor point, but there should probably be a qualifier there. At the career level an 8 PER player seems slightly more likely to be seen as one to be credited for adding some wins, but there are sizable chunk seen as taking away wins (http://bkref.com/tiny/H3rim). Looking at it on a season level (with a higher minutes requirement) the negative player chunk is smaller but still certainly non-neglibiable.

So WS does tend to think that an equivalent standard player at 8 PER should/will (typically) be contributing wins - it's just not universal and perhaps not too far off the tipping point where an equivalent to such a PER will be negative (by these type of crude studies somewhere between 6 and 7 PER).

I was more speaking to how PER works and comparing to WS. Not saying PER and WS correlate as I think you're describing.

I was sort of guessing that might be the case. Nevertheless it potentially comes across as WS allows you to be infinitely bad and still describes you as contributing wins - and in reality whilst your broad point, that unlike BPM and PER which set average as their baseline, or their respective per minute cousins VORP and EWA which set a "replacement level" below average but not 0 win level, which is where the Win Shares metrics sets its bar (i.e. considerably lower), stands - Win Shares does have the capacity for negative wins and it's not implausible that an 8 PER player would be in the negative range.

As before I think it either needed the caveat, or else at least the use of a higher PER (say 10) in the example - with the former offering a fuller picture.

[dhsilv2]

A minor point, but there should probably be a qualifier there. At the career level an 8 PER player seems slightly more likely to be seen as one to be credited for adding some wins, but there are sizable chunk seen as taking away wins (http://bkref.com/tiny/H3rim). Looking at it on a season level (with a higher minutes requirement) the negative player chunk is smaller but still certainly non-neglibiable.

So WS does tend to think that an equivalent standard player at 8 PER should/will (typically) be contributing wins - it's just not universal and perhaps not too far off the tipping point where an equivalent to such a PER will be negative (by these type of crude studies somewhere between 6 and 7 PER).

I was more speaking to how PER works and comparing to WS. Not saying PER and WS correlate as I think you're describing.

I was sort of guessing that might be the case. Nevertheless it potentially comes across as WS allows you to be infinitely bad and still describes you as contributing wins - and in reality whilst your broad point, that unlike BPM and PER which set average as their baseline, or their respective per minute cousins VORP and EWA which set a "replacement level" below average but not 0 win level, which is where the Win Shares metrics sets its bar (i.e. considerably lower), stands - Win Shares does have the capacity for negative wins and it's not implausible that an 8 PER player would be in the negative range.

As before I think it either needed the caveat, or else at least the use of a higher PER (say 10) in the example - with the former offering a fuller picture.

John's formula I believe is 10.5 on average for his 0 wins factor (varies by position). But again his view is that is the replacement level, while WS affords players below that threshold to be valuable players. Maybe 8 was too low, but the number is below 10, and I did open with I was using broad strokes, not exact math.