OT: What are the “Alabama” of sports stats

[Lafayette]

My vote is for

1. +\-
2. WAB
3. WAR

1 loses because it penalizes substitutes at free throws when they come in with the other team shooting and similarly penalizes for what the other 9 players do.

2. Does anyone really think the 45th NET team had a 25% chance of beating Michigan or Duke on a neutral court?

3. Is based on an AAA player if baseball but ignore the next man up could be marginally worse or better than the player they are replacing.

 

[auror]

My vote is for

1. +\-
2. WAB
3. WAR

1 loses because it penalizes substitutes at free throws when they come in with the other team shooting and similarly penalizes for what the other 9 players do.

2. Does anyone really think the 45th NET team had a 25% chance of beating Michigan or Duke on a neutral court?

3. Is based on an AAA player if baseball but ignore the next man up could be marginally worse or better than the player they are replacing.

#2 The NET gives a bubble team.a 10% chance to beat Michigan in a neutral (and thus 0.9 if you win), not 25%.

 

[Lafayette]

#2 The NET gives a bubble team.a 10% chance to beat Michigan in a neutral (and thus 0.9 if you win), not 25%.

Saw Norlander explain WAB on CBS Sports and said it’s 25% but now reading it’s game dependent eg our playing @Kansas would be lower pct likely than say Butler playing us at home for VCU the current bubble team

 

[EnochRoot]

It's my understanding that + / - in basketball is useful in terms of season-long data, but pretty much useless in game-level statistics.

Case in point. On 2/21, the Spurs beat the Kings 139-122. Stephon Castle's stats: In 25 minutes he had 18 pts (6-9 fg, 6-8 ft), 8 reb, 3 assists, 2 turnovers, 1 block, 2 fouls.

The +/- rated him as Zero. And that's because the score differential remained neutral while he was on the floor. There are way, way too many things outside the player's control if you're going to view that stat on the game level. The +/- to date though is quite handy.

 

[Waquoit]

My vote is for

1. +\-
2. WAB
3. WAR

1 loses because it penalizes substitutes at free throws when they come in with the other team shooting and similarly penalizes for what the other 9 players do.

2. Does anyone really think the 45th NET team had a 25% chance of beating Michigan or Duke on a neutral court?

3. Is based on an AAA player if baseball but ignore the next man up could be marginally worse or better than the player they are replacing.

3. Is good for absolutely nothing.

 

[auror]

Plain, single game plus/minus is pretty bad.

Plus/minus you need to adjust for teammates, then adjust for opponents and location. Then get a bigger sample, preferably multi-year. Maybe add a box score prior. Then you get the useful versions: RAPM, BPR, EPM, etc.

 

[husky429]

I've always like RAPM.

 

[QDOG5]

Bitd I really liked margin of victory as a predictor in the tourney. Calhoun and Roy Williams teams would run teams to death. Ben Howland teams would play games in the low 60s and lose in the round of 32. Happened to Tony Bennett one year. He figured it out the next year.