Over and Underachievers
Taking a look at the teams that missed their expected win total by four or more games in 2023-24.
Many years ago, Bill James noted that a baseball team’s true strength could generally be measured more accurately by looking at its runs scored and runs allowed rather than its win-loss percentage. That is, he found that one could predict future win-loss percentage more accurately by using the previous season’s runs scored and runs allowed rather than the previous season’s win-loss percentage.
The same holds true in basketball: a team’s point differential is a better indicator of its true strength than its win-loss percentage. This information can be used to find teams that either greatly underachieved (i.e., had a much lower win-loss percentage than would have been expected based on their point differential) or teams that greatly overachieved (i.e., had a much higher win-loss percentage that would have been expected based on their point differential).
Last year, I outlined how to convert a team’s average point differential into an expected win total:
For example, the 2023-24 LA Clippers had an average point differential of plus-3.3 PPG in 82 games. Their expected win total is:
The Clippers actually won 51 games, so the formula was almost right on. If this process is repeated for every team in NBA history, you’ll get a root-mean-square error of 3.1 wins* (for reference, last season’s figure was 2.8 wins).
* In seasons where fewer than 82 games were played, I prorated both wins and expected wins to an 82-game schedule.
For teams that missed their expected win total by at least four wins, I made a prediction for the following season:
If the team won more games than expected, then I predicted they would win fewer games the following season (i.e., notable overachievers will fall back).
If the team won fewer games than expected, then I predicted they would win more games the following season (i.e., notable underachievers will improve).
Among the 331 teams that missed their expectation by four or more wins and played the following season:
A total of 166 went over their expected win total, with 111 of them (66.9%) winning fewer games the next season.
A total of 165 went under their expected win total, with 106 of them (64.2%) winning more games the next season.
Overall, 217 of these “extreme” teams (65.6%) matched their prediction and 114 (34.4%) did not. Considering that no other factors were taken into account — most notably roster changes — that seems like a noteworthy result.
Let’s take a look at the five teams that missed their expected win total by four or more games in 2023-24, and make some educated guesses about their prospects for the upcoming season.
Boston Celtics
Actual: 64 wins
Expected: 68.1 wins
Difference: minus-4.1 wins
Naive Prediction: more than 64 wins in 2024-25
The Celtics average point differential last season was plus-11.3 PPG, the fifth-highest such figure in NBA history. That figure was padded by 10 wins of 30 or more points, tying them with the 1971-72 Milwaukee Bucks for the most such wins in league history.
Keep reading with a 7-day free trial
Subscribe to Statitudes to keep reading this post and get 7 days of free access to the full post archives.