Traditional statistics tell us what a player recorded.
Points. Assists. Rebounds. Steals. Blocks.
But coaches usually want to answer a harder question:
How much did this player actually help his team play better basketball?
That is the problem Estimated Plus-Minus tries to address.
On Dunks & Threes, the current EPM model is forward-looking and predictive, while the Actual section presents season-specific impact and statistics from the selected season. For the 2025–26 season, it separates player impact into offensive, defensive and overall EPM.
For coaches and players, that makes Actual EPM especially useful when reviewing what happened during a season and trying to understand why.
What Is Actual EPM?
Actual EPM can be read as an estimate of a player’s impact relative to league-average level, expressed in points per 100 possessions.
It has two main sides:
| Component | Basketball meaning |
|---|---|
| OFF EPM | Estimated offensive impact |
| DEF EPM | Estimated defensive impact |
| EPM | Combined overall impact |
VISUAL 1 — THE BASIC IDEA
OFF EPM + DEF EPM = TOTAL EPM
Offense → scoring, creation, efficiency, spacing, passing, ball security
Defense → rim protection, disruption, positioning, rebounding, team defensive impact
Total EPM → estimated overall effect on team performance
This is the first concept a coach should understand:
EPM is not asking how many statistics a player accumulated. It is trying to estimate how much those actions affected winning possessions.
Actual EPM vs Expected EPM
This distinction is important because Dunks & Threes changed the EPM framework.
The site’s main EPM model is now predictive. It uses historical information and what Dunks & Threes calls Estimated Skills to estimate the player’s underlying and future impact. The Actual page, by contrast, is labeled as the player’s actual impact and stats for the selected season.
VISUAL 2 — ACTUAL vs EXPECTED
| ACTUAL EPM | EXPECTED EPM |
|---|---|
| Looks backward at the selected season | Looks forward |
| What impact was attributed to the player? | What impact should we expect? |
| Useful for season review | Useful for projection |
| Retrospective evaluation | Predictive evaluation |
A simple coaching interpretation:
Actual EPM = What happened?
Expected EPM = What should we expect next?
Those are related questions, but they are not the same question.
Why Not Just Use Plus-Minus?
Suppose Player A finishes a game at +14 and Player B finishes at −6.
That does not automatically mean Player A played better.
Player A may have shared almost all of his minutes with four starters.
Player B may have led a weak second unit.
One may have faced the opponent’s best lineup. The other may have played against reserves.
Opponent three-point shooting can also swing dramatically during relatively small samples.
This is why raw plus-minus is noisy.
Dunks & Threes explains that EPM combines a Statistical Plus-Minus component with Regularized Adjusted Plus-Minus. RAPM attempts to control for who a player plays with and against, while the statistical component helps stabilize the estimate using measurable player information.
In coaching language:
Raw plus-minus asks what happened while you were on the floor.
Impact modelling tries to estimate how much of it was connected to you.
That is a much more useful distinction.
How Should a Coach Read the Number?
Do not overreact to decimals.
The difference between +3.2 and +3.4 should rarely change a basketball evaluation.
Think in impact bands.
VISUAL 3 — PRACTICAL COURT DECODE SCALE
| Actual EPM | Practical interpretation |
|---|---|
| +6 or higher | Elite / MVP-level impact |
| +3 to +6 | Major positive / All-Star-level impact |
| +1 to +3 | Strong positive contributor |
| 0 to +1 | Around average to slightly positive |
| −1 to 0 | Slightly below average |
| −3 to −1 | Clear negative impact |
| Below −3 | Very negative / replacement-level territory |
Important: these are Court Decode interpretation bands for easier coaching communication, not official Dunks & Threes classifications.
The more important idea is distance from league-average impact.
A +5 player is not simply someone with better box-score numbers than a +1 player.
The model is indicating a much stronger influence on possessions.
Offensive EPM Is More Than Scoring
One of the easiest mistakes is to read OFF EPM as a scoring metric.
It is not.
Two players can both average 20 points per game and have very different offensive value.
One may score 20 while requiring a huge number of possessions, stopping ball movement and producing turnovers.
Another may score 20 efficiently while also creating advantages, forcing rotations and improving the shots available to teammates.
High-level offensive impact can come from scoring efficiency, rim pressure, passing, advantage creation, spacing, offensive rebounding, ball security and off-ball value.
This is why the better question is not:
How many points did he score?
It is:
What happened to our offense because he was involved in the possession?
Defensive EPM Needs More Context
Defense is even harder to quantify.
A defender can execute a perfect possession without recording a steal or block.
He may:
tag the roller,
deny an entry pass,
force the ball away from the middle,
rotate as the low man,
switch correctly,
protect the rim without allowing a shot,
or finish the possession by boxing out.
Traditional box scores capture only part of this.
EPM therefore benefits from the adjusted plus-minus component and from information beyond basic box-score totals. Dunks & Threes also notes that defensive statistics remain more difficult to capture than offensive ones.
For coaches, that leads to an important rule:
DEF EPM should point you toward film. It should not replace film.
If a player grades extremely well defensively, ask why.
Is it rim protection?
Point-of-attack defense?
Switchability?
Rotations?
Rebounding?
Turnover creation?
Scheme fit?
The metric identifies the signal.
Video identifies the mechanism.
Real NBA Example: Different Paths to Elite Impact
The 2025–26 Actual EPM table provides a useful example.
Shai Gilgeous-Alexander is listed at approximately +7.3 OFF EPM, +2.0 DEF EPM and +9.3 total EPM.
Victor Wembanyama is at roughly +4.3 offense, +4.4 defense and +8.7 total.
Nikola Jokic is at approximately +7.5 offense, +0.7 defense and +8.1 total EPM.
VISUAL 4 — DIFFERENT WAYS TO CREATE ELITE IMPACT
Shai Gilgeous-Alexander
OFF +7.3 ██████████████
DEF +2.0 ████
TOTAL +9.3
Victor Wembanyama
OFF +4.3 █████████
DEF +4.4 █████████
TOTAL +8.7
Nikola Jokic
OFF +7.5 ███████████████
DEF +0.7 █
TOTAL +8.1
The key coaching lesson is not simply that all three are elite.
It is that their impact profiles are different.
Jokic produces extraordinary value primarily through offense.
Wembanyama produces unusually balanced two-way impact.
Gilgeous-Alexander combines dominant offensive value with strong positive defensive influence.
That tells us much more than comparing points per game.
EPM Is a Rate Metric — Playing Time Still Matters
Another important distinction:
Impact rate is not the same thing as total season value.
A player can post an excellent EPM over 400 minutes.
Another can produce slightly lower impact but sustain it for 2,700 minutes.
Those seasons should not automatically be considered equivalent.
Dunks & Threes also lists Estimated Wins (EW) alongside EPM, providing a way of moving from impact rate toward cumulative value.
For practical player evaluation, therefore, think:
Impact × Minutes × Role × Availability
not simply:
Who has the highest EPM?
What Actual EPM Can Reveal That the Box Score Misses
Imagine a player averaging:
17 PTS | 5 REB | 4 AST
The numbers look productive.
But his impact remains around league average.
That should create a question.
Maybe the player requires too many possessions.
Maybe he creates little advantage.
Maybe he turns the ball over under pressure.
Maybe his spacing is weak without the ball.
Maybe the defense targets him.
Maybe his individual production does not improve the possessions of the other four players.
Now consider another player:
9 PTS | 5 REB | 3 AST
but with strong positive impact.
Traditional statistics may be missing valuable contributions such as screening, spacing, quick decisions, secondary creation, defensive versatility, transition defense or rim deterrence.
This is where an impact metric becomes genuinely useful.
It helps identify players whose box-score profile and possession-level influence tell different stories.
How Coaches Should Use Actual EPM
VISUAL 5 — THE COURT DECODE WORKFLOW
1 — IDENTIFY
Find a player whose EPM does not match your first impression from the box score.
↓
2 — DECOMPOSE
Separate OFF EPM from DEF EPM.
↓
3 — INVESTIGATE
Check role, lineups, matchups, usage and film.
↓
4 — EXPLAIN
Determine which basketball actions are producing the impact.
↓
5 — APPLY
Turn the finding into a rotation, development, scouting or lineup decision.
This is much more useful than treating EPM as a leaderboard.
How Players Should Use It
Players should not ask:
“How do I increase my EPM?”
The better question is:
“How do I create more winning possessions?”
For a guard, improvement may come from creating advantages without adding turnovers.
For a wing, it may come through shooting gravity, closeout attacks, faster decisions and defensive versatility.
For a big, value may come from screen quality, short-roll decision-making, offensive rebounding, rim protection and coverage execution.
Many of those actions may not dramatically change points, rebounds or assists.
But they can change possessions.
That is the point.
What Actual EPM Cannot Tell You
No all-in-one metric can fully answer whether a player fits another team’s system, who should take the final shot, who should defend a specific matchup, which coverage should be used in a playoff series or how leadership affects a group.
Basketball context still matters.
Role matters.
Scheme matters.
Spacing matters.
Teammates matter.
Matchups matter.
Sample size matters.
And even sophisticated models remain estimates rather than direct measurements of individual basketball value.
That is why an EPM number should never be the final sentence of an evaluation.
It should be the beginning of a better question.
The Court Decode Takeaway
Analytics should start questions, not end them.
Actual EPM is useful because it pushes analysis beyond the traditional box score.
It asks us to think about possession impact, not just accumulated production.
For coaches, it can identify players who may be contributing more—or less—than their conventional statistics suggest.
For players, it reinforces an important principle:
Your value is not only what appears next to your name in the box score.
It is also what happens to the possession because you are on the floor.
The strongest evaluation process remains:
DATA → QUESTION → FILM → CONTEXT → DECISION
That is where analytics becomes basketball.
Source and methodology: Dunks & Threes, Estimated Plus-Minus methodology and 2025–26 Actual EPM table. EPM uses an SPM component as prior information within a RAPM framework; the current predictive version also incorporates Estimated Skills, historical information and tracking-related data.
