Poor man's player efficiency rating

[bigfive]

If you add up a player's points rebounds steals blocks and assists, and then subtract all missed shots turnovers and fouls committed, then divide by the number of minutes played, then multiply by 40, you get a rough approximation of how well each player contributes when they are playing. Here are the numbers so far this year, with the caveat that Gandy's is based on an inadequate sample of 4 minutes. Also, there are many contributions not measured in the box score. That said, these are the numbers so far this year: Sarah 44.35, Azzi 25.26, Blanca 20.00, Gandy 20.00, Allie 19.86, Serah 19.24, KK 18.12, Jana 16.44, Kelis 16.14, Kayleigh 16.13, Ice 13.91, Caroline 13.71, Ayanna 13.33, Ashlynn 11.43.

The actual gross number of net positive contributions per player: Sarah 265, Azzi 156, KK 101, Serah 76, Kayleigh 73, Allie 71, Ashlynn 60, Blanca 58, Jana 37, Caroline 24, Kelis 23, Ayanna 13, Ice 8, Gandy 2.

 

[WiseWillie]

If you add up a player's points rebounds steals blocks and assists, and then subtract all missed shots turnovers and fouls committed, then divide by the number of minutes played, then multiply by 40, you get a rough approximation of how well each player contributes when they are playing. Here are the numbers so far this year, with the caveat that Gandy's is based on an inadequate sample of 4 minutes. Also, there are many contributions not measured in the box score. That said, these are the numbers so far this year: Sarah 44.35, Azzi 25.26, Blanca 20.00, Gandy 20.00, Allie 19.86, Serah 19.24, KK 18.12, Jana 16.44, Kelis 16.14, Kayleigh 16.13, Ice 13.91, Caroline 13.71, Ayanna 13.33, Ashlynn 11.43.

The actual gross number of net positive contributions per player: Sarah 265, Azzi 156, KK 101, Serah 76, Kayleigh 73, Allie 71, Ashlynn 60, Blanca 58, Jana 37, Caroline 24, Kelis 23, Ayanna 13, Ice 8, Gandy 2.

Interesting #'s, bigfive. The drop off from Sarah to the others is certainly notable.

Do you know if these derived quantities of net positive contribution (or NPC) per player are used in any manner whatsoever for choosing the NPOY or is it not feasible since it would require a much longer set of calculations and probabilities based on a team's SOS and so many other factors to make it unworkable or meaningless?

 

[bigfive]

I don’t know what the different voters use as criteria, although choosing JuJu Watkins instead of Sarah or Paige last year suggests they just look at points per game and not efficiency, since both Sarah and Paige were much better in official player efficiency rating and win shares.

 

[Gael]

@bigfive, your stat is similar to Player Efficiency Rating (PER), which is the most commonly available stat designed to be a per-minute, pace-adjusted measure of a player's productivity for every minute they are on the court. Via weighed formulas, PER sums each player's positive accomplishments (2FGs, 3FGs, FTs, assists, rebounds, blocks, steals) and subtracts the negative ones (missed FGs, missed FTs, TOs, PFs).

PER is primarily a measure of pace adjusted offensive productivity for every minute a player is on the court, and is weaker on incorporating all-around defensive contributions.


Here are UConn's PERs after nine games as calculated by Sports-Reference:

Rk	Player	Pos	G	GS	MP	PER ▼
1	Sarah Strong	F	9	9	239	49.2
2	Azzi Fudd	G	9	9	248	33.9
3	Blanca Quinonez	G	7	0	117	26.6
4	Gandy Malou-Mamel	C	2	0	4	25.1
5	Allie Ziebell	G	9	0	144	24.7
6	Serah Williams	F	9	9	160	23.1
7	KK Arnold	G	9	9	220	23.0
8	Kelis Fisher	G	7	0	57	22.7
9	Jana El Alfy	C	9	0	90	21.6
10	Kayleigh Heckel	G	9	0	182	20.9
11	Ayanna Patterson	F	7	0	40	17.4
12	Ashlynn Shade	G	9	9	206	15.9
13	Ice Brady	F	2	0	23	14.1
14	Caroline Ducharme	G	7	0	71	11.6

Here are the national leaders in PER as of today according to Sport-Reference's implicit cutoffs:

1.	Audi Crooks Iowa State	55.9
2.	Sarah Strong UConn	49.2
3.	MiLaysia Fulwiley LSU	45.2
4.	Gracie Merkle Penn State	45.0
5.	Brooklyn Meyer South Dakota State	42.1
6.	Mia Nicastro Western Illinois	41.7
7.	Hannah Hidalgo Notre Dame	41.1
8.	Raegan Beers Oklahoma	40.2
9.	Mikayla Blakes Vanderbilt	39.6
10.	Avery Koenen North Dakota State	38.9

 

[WiseWillie]

The problem as I see with the PER it is that the metric developed by John Hollinger does not incorporate opponent-specific data such as defensive rating of a player or the opponent's overall team quality. It essentially considers that every player faces the same average quality of competition and treats players on all teams as being equal even though their team's SOS can be decidedly different. For use in the NBA where teams play other teams multiple times throughout a season the SOS is a lot less critical and the PER might be a better metric than use in the evaluation of college play.

 

[Gael]

The problem as I see with the PER it is that the metric developed by John Hollinger does not incorporate opponent-specific data such as defensive rating of a player or the opponent's overall team quality. It essentially considers that every player faces the same average quality of competition and treats players on all teams as being equal even though their team's SOS can be decidedly different.

WiseWillie, to the extent your criticisms make sense on an NCAA-wide basis, they have no relevance to the comparative PERs on the UConn team, which is the subject of this thread. Every player on the UConn team has faced the same opponents, so the SOS of opponents is irrelevant to intra-team PER comparisons. I threw in the NCAA-wide list only because I thought some folks here would be interested where Sarah Strong ranked nationally on PER at this early stage.

What does distort intra-team PERs are minor minute players such as Malou-Mamel, Patterson and Brady. Like all stats and rankings, PER calculations will become more accurate and informative as more games are played and more playing times are accumulated.

 

[bigfive]

It is true that the deep subs may play against weaker players late in the game, which can improve their numbers