Methodology

HOW NET IMPACT WORKS

Net Impact measures how much value a player creates and destroys relative to what an average player would produce in the same minutes. Every action is valued by its effect on possessions and points.

Every NBA possession is worth approximately 1.12 points. Every action a player takes either creates or destroys value relative to that baseline.

Net Impact sums all micro-values across a game, then asks: did this player produce more or less than what an average player would have in the same minutes?

Every offensive action is valued by its possession-economics impact.

Points +1.00 Face value — a point is a point
Off. Rebound +1.344 Creates a new possession at a higher conversion rate
Assist +0.50 Shot quality upgrade + playmaking value
Screen Assist +0.30 The screen that frees a teammate for an open shot
Secondary Assist +0.25 The hockey assist — pass before the pass
Missed 2PT -0.784 70% of the time the miss ends the possession
Missed 3PT -0.874 Long rebounds are harder to recover
Missed FT -1.00 Full opportunity cost — uncontested shot
Turnover -1.95 60% live-ball TOs give opponent fast break

Box score outcomes combined with tracking data that captures defensive actions beyond traditional stats.

Box Score

Def. Rebound +0.30 Prevents opponent offensive rebound opportunity
Steal +2.30 Two-possession swing — opponent loss + transition opportunity
Block +0.90 Possession ended + deterrent for future shots
Foul -0.75 Probability of sending opponent to line + accumulation risk

Tracking

Rim Protection Variable Suppression of opponent FG% at the rim vs league avg 66%
3PT Closeout +0.35/ea Most valued perimeter defensive action — All-Defense guards rank 90th+ pctl
Forced TOs Variable Turnovers by your assignment beyond your own steals
Box-Out Credit +0.30/ea Box-outs enabling a teammate rebound — invisible effort

Effort and positioning stats that separate high-motor players from passive ones.

Contested Shot +0.20 Contesting reduces opponent FG% by ~10%
Deflection +0.65 #1 reliable perimeter defense signal — ~20% chance of forcing TO
Charge Drawn +2.70 Turnover value + offensive foul on opponent
Loose Ball +0.60 A 50/50 ball won — similar to an offensive rebound

Five data-driven adjustment layers that capture value the box score misses.

1. Per-Event Classification

Instead of fixed values, each event is classified from play-by-play data. A live-ball steal (+2.37) is worth more than a dead-ball steal (+2.24). A shooting foul (-1.56) costs more than a non-shooting foul (-0.30).

2. Matchup-Level Defense

Measures how opponents shoot when guarded by each player vs their season average. Applied at 50% weight to avoid double-counting.

3. Clutch & Garbage Time

Actions in clutch time (last 5 min, within 5 points) get a bonus. Garbage time (20+ point margin in 4th) is discounted.

4. Shot Selection Quality

Each shot's expected value based on league-wide FG% for that shot type vs 1.12 PPP. Measures the decision, not the result.

5. Offensive Fouls Drawn

Full picture of foul-drawing from Basketball Reference PBP data. Toumani Camara leads the NBA with 102 offensive fouls drawn — only 18 were credited as charges. The other 84 (+2.91/game) were invisible until this layer.

Raw impact alone is meaningless — the question is: did you produce more or less than what any average player would have in your minutes?

Expected = your minutes × (all players' raw ÷ 480)
Net Impact = your raw production − expected

It uses game-specific context. A defensive grind has a lower baseline than a 130-125 shootout. The total impact across all players sums to exactly zero — a true zero-sum measure.

The baseline implicitly adjusts for pace. Fast games have higher baselines. Net Impact is pace-neutral without requiring explicit correction.

+8.0

Elite. All-Star caliber. Jokic and SGA average ~+14.

+3.0

Strong. Quality starter or elite role player.

0.0

Average. Being average at 35 minutes against NBA competition is hard.

-3.0

Below average. High usage with low efficiency, or a role mismatch.

Rim protection anchors an entire defense in ways perimeter defense can't. Teams can scheme around a great perimeter defender via screens. They can't move the opposing center away from the rim.

Offensive rebounds are worth +1.344 each — creating an entirely new possession. 4 OREBs per game = ~5.4 points of value from rebounding alone.

What the model can't measure: Floor spacing, playmaking gravity, help defense rotations, defensive switching versatility. These benefit guards and wings but are invisible to every statistical model.

We chose to measure what's measurable and be transparent about what isn't. See the Leaderboard for position-adjusted rankings.

Captures well

  • + Scoring efficiency at volume
  • + Shot selection quality
  • + Turnover cost by type
  • + Rim protection & interior D
  • + Hustle plays & deflections
  • + Matchup-level defense
  • + Offensive fouls drawn
  • + Clutch performance

Cannot measure

  • - Floor spacing / gravity
  • - Help defense rotations
  • - Switching versatility
  • - Playmaking gravity
  • - Pre-possession disruption
  • - Behavioral deterrence

Why does my player have negative impact despite great stats?

Usually turnovers. 25 points + 5 TOs = -9.75 from turnovers alone. The box score looks great; Net Impact shows they consumed heavy minutes at an average rate.

Does it penalize high-minute players?

No — 30+ minute players average +1.73 impact. But variance is higher (stdev 8.9 vs 2.6 for bench players). High minutes = high stakes.

Does pace affect Net Impact?

Not directly. The game-average baseline adjusts implicitly — fast games have higher baselines. Net Impact is effectively pace-neutral.

Explore the model

Leaderboard Trail Blazers Nuggets