Hustle Plus-Minus
How many points per 100 possessions a player adds through hustle: winning extra possessions, ending the other team's, and the small effort plays in between. Every play is valued by what it is worth on the scoreboard. Plays we can't value directly (deflections, screens, loose balls, contests) get the weight the data says they earn in winning, and most earn very little. Positive = more hustle value than an average player.
Leaderboard
Click a player to open his full breakdown. Column values are each part's points per 100 possessions.Offensive vs defensive hustle
Across: O-HPM. Up: D-HPM. Both in points per 100 possessions vs an average player.
Filled dots are clearly above or below average (more than twice their noise range). Hollow dots could still be noise.
Click any dot to open that player's breakdown.
Kings of
| # | Player | Value |
|---|
Method
Data
Play-by-play from NBA.com's live data feed for 2020-21 through the current season: about 1.46 million possessions, each tagged with the ten players on the floor (99.9% complete). Tracking stats (deflections, loose balls, screen assists, contests, box-outs, contested rebounds, distance and speed) come from NBA.com's player-tracking hustle data, pulled month by month. Charges match between the two sources player by player (r = 0.995).
Each season is scaled to the six-season league rate before seasons are blended, because some tracking stats changed how they are counted. League deflections rose 24% from 2021-22 to 2025-26 while contests fell 19%. Charges fell 32% after the rule change.
Points value of each play
Measured from possessions in 2023-24 to 2025-26. League average: points per possession.
Rebounding impact
oREB and dREB Impact are lineup-adjusted (RAPM): a ridge regression of each possession's offensive rebound rate on all ten players, controlling for score margin (rubberband) and home court. Changes in rebound rate are converted to points: one percentage point of team offensive rebound rate is worth points per 100 possessions, since an offensive rebound leads to points on average.
Learned weights
For stats without a direct points value, the weight is how much each one predicts lineup-adjusted impact beyond steals, blocks, charges and rebounding (controlling for role). Weights are pulled toward zero by their uncertainty. Points per 100 possessions, per one extra of the stat per 100:
Deflections, contests and box-outs add almost nothing once steals and rebounding are counted: the deflections that matter become steals, and box-outs show up in dREB Impact. They stay on the page as their own stats and in the Kings of section.
Does it repeat?
Year-to-year correlation of single-season scores, players with 2,000+ possessions in both seasons:
Noise
Each part's noise range comes from its sample size (counting stats) or the regression's uncertainty (rebounding impact). Likely real means HPM is more than twice its noise range away from zero.
Teams
Team HPM is the possession-weighted hustle of the players who played for that team in the selected view, times five players on the floor. In the Current view that is each team's present roster, time-weighted.
Not covered
- Reach-in fouls aren't labeled in play-by-play; "other non-shooting fouls" is the closest available stand-in.
- Transition defense, closeouts and communication don't show up in either data source.
- Tracking stats are only available as player totals, so they can't be lineup-adjusted.