NBA Prop Hit Rate vs. Model Probability: How to Compare Them
A hit rate describes a sample of past games. A model probability is an estimate derived from a model. Both can help you inspect a prop, but neither guarantees the next result.
What a hit rate actually counts
If a player exceeded 24.5 points in 6 of the last 10 games, the hit rate against that threshold is 60%. This answers a narrow question about those ten performances. It does not account for every difference between those games and the next matchup.
One additional hit changes a 10-game rate by 10 percentage points; it changes a 5-game rate by 20 points. Always read the sample count alongside the percentage. Head-to-head samples can be particularly small or span different player roles.
Why the model percentage can be different
The player-prop analysis uses a smoothed historical distribution, called kernel density estimation, to estimate probability on either side of a line. A distribution uses the locations of past results rather than only counting whether each cleared the threshold.
Consider two samples that each went over in 6 of 10 games. One clusters close to the line, while the other includes several much higher scores. The hit rates match, but their estimated distributions can differ. Smoothing choices and limited samples also affect the output.
A more precise-looking percentage is not proof of a better forecast. Calibration requires comparing probabilities with later outcomes across many predictions, ideally using data not used to fit the model. This page does not claim independently validated forecast accuracy.
Keep current-line and historical-line evidence separate
A current-line hit rate asks how often past performances would have cleared the displayed line. A historical-line hit rate asks how often the player cleared the line actually recorded for each past game.
Historical lines move with circumstances. A strong record against older, lower lines does not establish value against a higher line today. Missing historical prices also limit what can be inferred about past profitability.
A high hit rate can still come at a poor price
Illustrative example: −150 odds require a 60% win probability to break even with no push. Even if you assumed a 60% past hit rate was a reliable forecast, that price would offer zero expected profit before any additional costs.
At −110 the break-even threshold is 52.38%. Comparing an estimated probability with the price is more informative than sorting only by hit rate, but the estimate can be wrong. Inspect the sample, distribution, game date, and available price together.
Put the explanation next to the data
Browse NBA player profiles for free season averages and recent game logs, including players without a prop line today.
Start with the free homepage prop example, then explore the player prop listings and expected value rankings. Locked previews show where subscription access is required.
Compare plans for full analysis and rankings. Check the features listed for each plan before subscribing.
Continue reading: NBA player props: lines, odds, and a worked EV example or how to interpret NBA referee trends.
Researching a role change? Read how minutes, usage and bench roles affect player prop samples.