Aging Curve
How fast do players decline, and does it depend on position? Every NBA player 26 or older with 500+ minutes in 2025-26, projected for 2026-27 with an 80% range. Ratings are LEBRON, a public all-in-one rating from BBall Index in points per 100 possessions: 0 is an average player, about -2 is replacement level, +2 or more is a clear starter. A Ghost Screen Analytics model, not an NBA stat. The tests were written before the model was run.
How to read it
- 2025-26
- The player's LEBRON last season.
- Projected
- Expected 2026-27 LEBRON. The dark tick on each range.
- Change
- Projected minus 2025-26. Green is a projected gain, red a projected drop.
- 80% range
- The span expected to hold for about four of every five players. It is wide on purpose. Young players swing a lot from year to year. The hollow circle is the 2025-26 rating.
Who is left out. The model only scores players who play 300+ minutes the next season. Players who retire or lose their minutes are not in the test, so real declines are bigger than shown.
The average age curve
| Age | All players | Guards | Forwards | Centers | Player-seasons |
|---|
Average change in LEBRON from this season to the next, by age that season. Ages under 19 or over 37 are grouped with 19 and 37. A dash means fewer than 15 player-seasons. Description only, not one of the seven bars.
The projections
Dark tick = projection. Hollow circle = 2025-26 rating. Light line = 80% range. Thin vertical line = average NBA player (0).
What it is built on
Sample: player-seasons tested, 2016-17 to 2025-26, from LEBRON ratings of players 26 or older with 500+ minutes, paired with the next season. Each test season used only earlier seasons for training. Inputs: current, offensive, defensive and last-year LEBRON, minutes, and age (with age squared and cubed) combined with position group. Regular season only.
The rules, code, data and results are in the aging-model folder on GitHub. Passing six bars on ten seasons is not proof the model will keep working.