APBRmetrics

TOV%
Turnover Percentage (available since the 1977-78 season in the NBA); the formula is 100 * TOV / (FGA + 0.44 * FTA + TOV). Turnover percentage is an estimate of turnovers per 100 plays.
from http://www.basketball-reference.com/about/glossary.html

Wouldn't it be a better stat to have it as a per individual possession used? Like say Player1 uses 10 possessions this game and turns the ball over 2 times, His Individual Turnover Rate would be 20%. Or even have it 100 * (TO * (Tm MP / 5)) / (MP * Opp Poss) just like Steal%.

Wouldn't that be a better indicator? insead of just TO/Possessions for TOV%

I believe it is implied that it is per 100 plays used by the player. So, if you see a player with 12% TOR, that's per 100 plays used by that player, not per 100 team plays.

EvanZ wrote:I believe it is implied that it is per 100 plays used by the player. So, if you see a player with 12% TOR, that's per 100 plays used by that player, not per 100 team plays.

They use the same formula for teams. And it is TOV%, not turnover rate. For individual players turnover rate makes more sense, because that incorporates assists, while TOV% does not. It makes no sense to include assists on a team basis, because those plays are already accounted for with the FGA.

Poldi now suggest that it is better to use possessions instead of plays. Meaning, we are subtracting ORB from the team data.

TOV% = 100*TOV/(FGA+0.44*FTA+TOV-ORB)

Seems not like a bad idea.

Offensive rebounds don't subtract linearly at the player level, which is why I think this is just an estimate. I could be wrong, but this is from BoP, no?

Using Chaikin's TO% might be better. 100 * TOV /[FGA + TOV + (FTA / (Tm FTA / Opp PF)) + (AST/0.17)]

It definitely makes more sense to have it at the player level (Tos / Player Possession) rather than the team level (TOs / Team Possessions).

Take the following:
Under Chaikin's TO%, if players' have 3 TO and 0 FGA their TO% would then be worse than 3 TO and 3 FGA.

samuel dalembert has a career TOV% of 17.2%, jose calderon 16.6%. sounds similar...

but per touch calderon has career turnovers per touch of just 3% (1 turnover for every 33 or so touches on offense), dalembert 12% (1 turnover for every 8-9 touches on offense)...

i prefer the latter (turnovers per touch) simply because i tend to think of whether on offense i would want the ball in the hands of a player who commits a turnover once every 33 touches versus the hands of a player who commits a turnover as often as once every 8 touches...