Why Rookie Sensations
Almost Always Fade
A rookie has a stunning first season. Sports writers call the next year a slump, or a curse. Nothing is cursed. Watch it happen to players who don't even exist yet.
Act one
Draft a rookie class
Twenty simulated players, each with a fixed hidden skill level, play two seasons. Their score each season is their skill plus a random dose of luck. Watch what happens to last year's top five, highlighted below.
Every player, skill level, and season is generated fresh in your browser. Nothing here is staged.
Act two
How much of this is luck?
Slide the balance between pure skill and pure luck and watch how strongly performance snaps back toward average from one season to the next.
At 50% luck, last year's top 5 are still above average, just far less extreme.
Nothing has to be wrong for things to average out
Francis Galton first noticed this in 1877, measuring the diameter of sweet pea seeds. The offspring of unusually large seeds were smaller than their parents, on average, and the offspring of unusually small seeds were larger. In 1886 he found the same pattern in people, showing that unusually tall parents tended to have somewhat shorter children, and unusually short parents somewhat taller ones. He called it "regression toward mediocrity", and the name regression to the mean stuck.
The mechanism is exactly what you just saw. Any result that mixes a stable trait with a dose of randomness will occasionally land on an extreme, and extremes almost always involve the randomness cooperating. The next measurement draws a fresh dose of randomness that has no reason to cooperate the same way again, so the result drifts back toward whatever the stable trait alone would produce. Nothing declined. The luck simply reset.
Daniel Kahneman told a version of this that changed how flight instructors were trained. Israeli Air Force instructors were convinced that praising a pilot after a great landing made the next landing worse, while yelling at a pilot after a bad one made the next landing better, so they gave up on praise entirely. What they were watching was regression to the mean: an unusually great landing was likely to be followed by a more ordinary one regardless of what anyone said, and the same was true in reverse for an unusually bad one. Criticism looked like it worked only because bad performances were already due to improve on their own.
- The "Sports Illustrated cover jinx": athletes tend to perform worse right after landing on the cover, not because of a curse, but because making the cover in the first place usually requires an unsustainably hot streak.
- Patients often seek treatment exactly when a chronic symptom is at its worst, and feel better afterward almost regardless of whether the treatment did anything, simply because the symptom was likely to ease up on its own.