Do players play better
in a contract year?
Every contract year since 1991, compared with the same player's own normal: the two seasons before and the two seasons after. Then the same test on every ordinary season, so only the part that's extra counts.
Watch all five versions of the model get built.
95% CI +0.11 to +0.37, 839 contract years. About 0.12 standard deviations: real, but small.
95% CI +1.7 to +3.3. Then 4.4 fewer games the season after signing.
Random ordinary seasons that beat the real contract-year bump (p ≈ 0.001).
of contract years, vs 2.1% of ordinary seasons. In a player's prime: 3.8%.
The method
His own normal
Each contract year is compared with the minutes-weighted average of the same player's two seasons before and two after (300+ minutes each), adjusted for age. Contract years need 500+ minutes.
Ordinary seasons, same test
Every non-contract season gets the exact same comparison. Ordinary seasons stick out from a player's normal by 2.0 PER (one standard deviation), so the bump only counts the part beyond that.
10,000 random tries
Each player's contract year is swapped for one of his own ordinary seasons, 10,000 times. In the prime-age sample only 9 of those fake versions matched the real bump.
Every contract year
PER in the contract year vs his normal. "SD" is how unusual the bump is compared with ordinary seasons; 2+ is an outlier. Search a name or tap a column to sort.