NBAValuation

Findings · performance through 2021-22 · salaries through 2023-24

How NBA pay tracks production

A salary benchmark built from public production data, compared with what players were paid that season and one and two seasons later. The results describe historical pay patterns. They do not measure a player's worth or prove that anyone was mispriced.

Finding 1 · descriptive

The most productive players earn far less in their first four seasons

Among each season's most productive third of players, those 0–3 years from their debut averaged 4.5% of the salary cap and those 8+ years in averaged 17.9%. The pattern fits rookie-scale and restricted free agency rules, but it is not proof of them. Contract status is not in the public data.

0%5%10%15%20%012345678910+Years since NBA debutBottom third, 0 years since debut: 2.3% of the cap (95% CI 2.1–2.6), n = 237Bottom third, 1 years since debut: 2.3% of the cap (95% CI 2.0–2.5), n = 184Bottom third, 2 years since debut: 2.8% of the cap (95% CI 2.5–3.1), n = 116Bottom third, 3 years since debut: 3.6% of the cap (95% CI 3.1–4.2), n = 95Bottom third, 4 years since debut: 6.2% of the cap (95% CI 5.1–7.5), n = 57Bottom third, 5 years since debut: 6.8% of the cap (95% CI 5.2–8.4), n = 45Bottom third, 6 years since debut: 7.2% of the cap (95% CI 5.6–8.9), n = 37Bottom third, 7 years since debut: 9.6% of the cap (95% CI 7.6–11.8), n = 39Bottom third, 8 years since debut: 6.5% of the cap (95% CI 4.9–8.3), n = 34Bottom third, 9 years since debut: 7.1% of the cap (95% CI 5.3–9.4), n = 36Bottom third, 10+ years since debut: 7.2% of the cap (95% CI 5.8–8.7), n = 139Bottom thirdMiddle third, 0 years since debut: 2.7% of the cap (95% CI 2.3–3.1), n = 109Middle third, 1 years since debut: 2.6% of the cap (95% CI 2.3–2.9), n = 134Middle third, 2 years since debut: 2.9% of the cap (95% CI 2.6–3.3), n = 125Middle third, 3 years since debut: 3.8% of the cap (95% CI 3.3–4.4), n = 97Middle third, 4 years since debut: 8.1% of the cap (95% CI 6.7–9.6), n = 78Middle third, 5 years since debut: 9.0% of the cap (95% CI 7.6–10.4), n = 84Middle third, 6 years since debut: 10.0% of the cap (95% CI 8.5–11.6), n = 73Middle third, 7 years since debut: 9.7% of the cap (95% CI 8.1–11.4), n = 58Middle third, 8 years since debut: 10.5% of the cap (95% CI 8.6–12.4), n = 45Middle third, 9 years since debut: 11.4% of the cap (95% CI 9.2–13.8), n = 45Middle third, 10+ years since debut: 10.0% of the cap (95% CI 8.5–11.6), n = 169Middle thirdTop third, 0 years since debut: 3.4% of the cap (95% CI 2.7–4.3), n = 30Top third, 1 years since debut: 3.5% of the cap (95% CI 3.0–4.1), n = 71Top third, 2 years since debut: 3.6% of the cap (95% CI 3.0–4.2), n = 111Top third, 3 years since debut: 6.3% of the cap (95% CI 5.4–7.2), n = 121Top third, 4 years since debut: 15.1% of the cap (95% CI 13.7–16.5), n = 115Top third, 5 years since debut: 14.5% of the cap (95% CI 13.1–16.1), n = 97Top third, 6 years since debut: 15.4% of the cap (95% CI 13.9–17.0), n = 90Top third, 7 years since debut: 16.2% of the cap (95% CI 14.5–17.8), n = 79Top third, 8 years since debut: 18.0% of the cap (95% CI 15.7–20.3), n = 70Top third, 9 years since debut: 19.2% of the cap (95% CI 16.8–21.6), n = 59Top third, 10+ years since debut: 17.5% of the cap (95% CI 14.6–20.3), n = 176Top third
  • Bottom third of WAR (within season)
  • Middle third of WAR (within season)
  • Top third of WAR (within season)

3,055 player-seasons, 2014-15 to 2021-22. Details

Finding 2 · association

At the same salary, a higher benchmark goes with higher pay next season, most of all early in careers

Holding current cap share, career stage and season fixed, one more point of cap in the benchmark goes with 0.45 more points next season on average (95% CI 0.37 to 0.55). Per point it is 1.78 for players 0–3 years in and 0.40 and 0.34 later. Early-career benchmarks vary less, so per standard deviation the contrast is smaller (2.1 against 1.5 and 1.4 points). The average leans on large pay changes: without the most influential rows it is 0.28, and the median response (post hoc) is 0.09.

0.00.51.01.52.0Later cap share per point of benchmarkAll playersAll players: 0.45 (95% CI 0.37 to 0.55); n = 1,6510-3 years since debut0-3 years since debut: 1.78 (95% CI 1.52 to 2.10); n = 8064-7 years since debut4-7 years since debut: 0.40 (95% CI 0.27 to 0.53); n = 4728+ years since debut8+ years since debut: 0.34 (95% CI 0.20 to 0.48); n = 373

1,651 player-seasons with an observed later salary; missing salaries are unknown, not zero. Details

Finding 3 · forecast

But it adds little to a one-season pay forecast beyond current salary and career stage

Adding the benchmark to a regression on current salary and career stage moves mean absolute error from 2.66 to 2.65 points of cap (difference −0.003, 95% CI −0.065 to 0.061). Simply repeating this season's cap share scores 2.33. That edge mostly reflects the loss function: refitted by median regression (post hoc), the best regression scores 2.30, and the benchmark lowers error by a small 0.035 points. The benchmark does reduce large misses (lower root mean squared error).

0123Mean absolute error, points of capRepeat current cap share2.33Current salary + career stage2.66+ salary benchmark2.65+ minutes and WAR2.61

1,325 forecasts made at the end of each season, trained only on salaries already known. Details

Finding 4 · model accuracy

The salary model is accurate on average and least reliable for the highest-paid players

Gradient boosting, statistically tied with random forest, misses held-out salaries by $3.82M on average, 17.3% less than linear regression. Its 80% range held 80.4% of held-out salaries overall, and 14 of 26 at 20–30% of the cap and 3 of 15 at 30% or more. Every miss in those tiers was a salary above the range.

Held-out MAE
$3.82M
Better than OLS
17.3%
Range held, 30%+ tier
3 / 15

2021-22 held-out season. Model Lab · Range diagnostics

Salary is published annual earnings from a research workbook. Its upstream permissions are not independently established, and it lists about 450 players a season, so a missing salary is unknown, not zero. Newer seasons are not covered: see Methodology & Data for the source audit.

Aggregate-only build

Model evidence, without player pages

This build publishes the analysis, the model's held-out evidence and the methodology. Player-level pages are left out because they need player-level source data. The repository's full local build adds them.

Held-out MAE
$3.82M
Coverage of the 80% range
80%
Seasons better than OLS
5 / 5
Valued player-seasons
1,903

Project overview & code Open the Model Lab Methodology & Data