Best historic (pre-pbp era) impact metric?

Moderator: Doctor MJ

jalengreen
Bench Warmer
Posts: 1,451
And1: 1,088
Joined: Aug 09, 2021
   

Best historic (pre-pbp era) impact metric? 

Post#1 » by jalengreen » Thu Sep 16, 2021 11:35 pm

I've heard different names like AuPM and WOWY and what not.. wondering if there's been an analysis on correlation with RAPM or something??
blabla
Sophomore
Posts: 156
And1: 76
Joined: May 23, 2012

Re: Best historic (pre-pbp era) impact metric? 

Post#2 » by blabla » Sat Sep 18, 2021 1:59 pm

WinShares is pre-pbp and probably better then PER. I don't think there were many other metrics than those
I don't think WOWY means what you think it means
jalengreen
Bench Warmer
Posts: 1,451
And1: 1,088
Joined: Aug 09, 2021
   

Re: Best historic (pre-pbp era) impact metric? 

Post#3 » by jalengreen » Sat Sep 18, 2021 7:04 pm

blabla wrote:WinShares is pre-pbp and probably better then PER. I don't think there were many other metrics than those
I don't think WOWY means what you think it means


WOWYR* apparently is what Ben Taylor calls it

https://backpicks.com/2017/11/17/part-iv-historical-impact-multiple-wowyr-studies/
ceoofkobefans
Junior
Posts: 400
And1: 209
Joined: Jun 27, 2021
   

Re: Best historic (pre-pbp era) impact metric? 

Post#4 » by ceoofkobefans » Sun Oct 17, 2021 11:28 pm

If you go past the PBP data almost all of the metrics you’ll be able to use are Box metrics which aren’t terrible but aren’t enough on their own to make complete analysis on a player although they can put you in the right direction (less so than data ball metrics can). The best box models in the pre data all era IMO are

RAPTOR
BP BPM
BPM
PIPM
WS (I use Per possession like one would for Box stats)

There is also WOWY and WOWYR. The simplest way to describe this is that it is Like On/Off NRTG but on a game level (instead of looking at how well a team plays when a player in on or off the court it looks at how well the team plays in games the player plays or doesn’t play in) This is a good way to measure impact since it is looking at raw impact and doesn’t have the limitations of box metrics. The biggest problem with WOWY is that it is incredibly hard to get a good sample size for most players (in single or multi year stretches) since you need at least ≈ 10 games missed and played to get any sort of conclusive data.

The only other stat there is which is AuPM which uses Harvey Pollock’s +/- data combined with a box model to estimate RAPM. For the entire NBA it goes back to 1994 and for 76ers players it goes back to 1977 (Dr J Moses and Chuck Data!)

These are all of the stats i would deem good pre databall era and I would rank them

AuPM
RAPTOR
WOWY/WOWYR
BP BPM
PIPM/BPM/WS/100

You can find

WS and BPM on basketballreference.com
RAPTOR on 538.com (the website has Modern RAPTOR but they have a link to a historical raptor Database)

WOWYR/WOWY/BP BPM/AuPM on backpicks.com (most of it is under a paywall but WOWYR is free)

PIPM is very hard to find but there are a few old PIPM databases on spreadsheets throughout the internet

Return to Statistical Analysis