Using our knowledge about player availability for every player and every game, we can adjust win probabilities in head-to-head matchups and the resulting win totals for every team.

For each player we can compute their expected minutes played by taking their nominal play time, the likelihood that they will play in any given game, and the way in which minutes are re-allocated when their teammates sit out.

The individual contributions of each player to the team’s likelihood of winning is combined with our estimate of how many minutes they will play. This adjustment is done on both sides of the ball. Using this we produce an updated win probability estimate for every game which reflects the probabilistic availability of every player on the court.

Check out the Clippers’ game-by-game win probabilities starting from Feb 26 – adjusted by potential player absences –for the remainder of the 2019-20 NBA season:

In this example, we look at this effect on Los Angeles Clippers for the last 24 games of their season. This shows that in general, the Clippers see a reduction in win probability relative to their nominal odds. This is because they have two superstars who regularly miss games in Paul George and Kawhi Leonard. Their adjusted odds take a very pronounced hit on back-to-back games where not only is the team at a disadvantage because of the fatigue associated with that scenario, but Leonard has very high probability of missing the game. Over the 24 games in this analysis, the nominal Clipper roster would be expected to win 18 games while the estimate which includes the Fansure player availability estimates predicts they will only with 16. In a densely packed Western Conference, those two wins could have big implications for playoff seeding.