A hit rate is the simplest signal in NBA prop betting: what fraction of recent games did a player clear a given threshold? It's intuitive, easy to calculate, and easy to act on. It's also one of the most misused statistics in prop betting, because the conditions under which a hit rate is genuinely informative are more specific than most bettors realise.
This guide covers the statistics of hit rates: what sample size you need before a hit rate is meaningful, how to balance recency against reliability, the four situations where hit rates are systematically misleading, and a reading checklist for evaluating any hit rate you encounter in the wild.
What is a hit rate?
A hit rate is the proportion of games in a given sample in which a player's actual stat cleared a specified threshold. If a player scored more than 20 points in 7 of his last 10 games, his 10-game hit rate at 20+ points is 70%.
Hit rates can be calculated against fixed thresholds (arbitrary numbers like 15+ or 20+ points) or against the live sportsbook line. Line-relative hit rates — where the threshold is set equal to tonight's posted prop line — are more useful for betting because they directly measure how often the player has beaten the market's expectation, not just an arbitrary level. This distinction matters enormously: a 70% hit rate at 15+ points is trivial if the player averages 22; a 70% hit rate at 20.5 points when the player averages 21 is genuinely interesting.
The sample size problem
The fundamental problem with hit rates is that the samples used to calculate them — typically 5 to 20 games — are far too small for conventional statistical significance. A player who hits 7/10 (70%) over a 10-game sample could just as easily have a true hit rate of 50%, with the 7/10 result being nothing more than a lucky run.
To understand how unreliable small samples are, consider the 95% confidence intervals around a sample hit rate. If a player hits 70% in a 10-game sample, the 95% confidence interval around that estimate spans roughly 40% to 90%. That's not a useful estimate — it tells you the true hit rate is almost certainly somewhere between 40% and 90%, which includes nearly every meaningful probability value.
Here is how confidence interval width varies with sample size, for a sample hit rate of 70%:
10 games: 95% CI ≈ 40%–90% (width: ~50 percentage points)
20 games: 95% CI ≈ 49%–85% (width: ~36 percentage points)
50 games: 95% CI ≈ 57%–81% (width: ~24 percentage points)
100 games: 95% CI ≈ 61%–78% (width: ~17 percentage points)
The implication is uncomfortable: you need 50+ games to get a confidence interval narrow enough to be confident you're above the 52% implied probability typical for near-even-money NBA props. Most bettors are working with 10-game or 20-game samples. Those samples can produce directional signals, but they cannot produce statistical certainty.
This doesn't mean small samples are worthless — it means they should be treated as hypotheses rather than conclusions, investigated rather than bet blindly.
The recency-reliability tradeoff
If larger samples are more reliable, why not always use the largest sample available — the full season or multi-season history? Because larger samples are less sensitive to recent changes.
An NBA player's situation changes throughout the season. Role changes, lineup reshuffles, injury returns, trade deadline moves, coaching changes — each of these alters the underlying distribution of outcomes. A player who moved from bench to starter in January has a statistical profile in January that is simply a different thing from his statistical profile in October. Using a season-long hit rate that spans both periods blends two different distributions into a single number that may not represent either accurately.
The optimal approach depends on what you're trying to measure. If you want to detect a recent role change or usage shift, use last-10 or last-5 data. If you want to verify that a recent trend is consistent with a longer-term pattern, compare last-10 to season. If you're looking for stable structural tendencies (back-to-back performance, road vs home splits, specific opponent matchups), use as much data as you have.
The practical heuristic: use last-10 as your primary window, verify against season when possible, and treat last-5 as an alert that something may have changed but requires investigation to confirm.
Four situations where hit rates lie
1. Schedule-driven hit rates
A player on a recent run of high hit rates may simply have been facing weak defensive opponents. If the last ten games included disproportionately many teams that rank poorly against this player's position, the hit rate reflects the schedule, not the player's underlying tendency. Check defensive quality of the opponents in the sample before drawing conclusions.
2. Lineup-driven hit rates
Hit rates calculated during a period when a key teammate was injured inflate the player's usage and counting stats. When the teammate returns, the hit rate is unlikely to sustain because the underlying usage driver has changed. Always check whether there are lineup differences between the sample period and tonight's game.
3. Pace-driven hit rates
A stretch of games against fast-paced opponents inflates counting stats across the board. Points, rebounds, and assists are all pace-sensitive. A player's elevated hit rate over a period that included five consecutive games against the league's fastest teams will partially revert when he returns to a normal pace schedule.
4. Sample selection bias
Hit rate tools often show you the players with the highest recent hit rates, not a random sample of all players. The fact that you're seeing a player at the top of a hit rate list partly reflects that he got lucky — many players who have been performing similarly in true probability terms just didn't happen to hit as consistently in the specific recent window. This is a form of survivorship bias. The players shown are the ones whose recent variance was favourable, which means their true hit rates are likely lower than the displayed rates suggest.
Hit rate vs implied probability
A hit rate only tells you something useful about value when compared to the implied probability of tonight's price. Three caveats apply.
Caveat 1: The hit rate sample and tonight's line may not be measuring the same thing. If last season's 10-game hit rate was calculated against lines of 18.5–20.5, and tonight's line is 22.5, the hit rate is not directly comparable. The sample's threshold needs to match or approximate tonight's line for the comparison to be valid.
Caveat 2: Sample hit rates are noisy estimates of true probability. Even a 70% sample hit rate might reflect a true probability of 55%. When comparing hit rate to implied probability, the confidence interval matters. A sample hit rate of 70% vs an implied probability of 52% sounds like an 18-point edge, but it might be a 3-point edge or a –10-point negative edge, given the uncertainty in the sample. Do not treat the raw gap as the edge.
Caveat 3: Both numbers can be wrong simultaneously. The hit rate might overestimate due to schedule effects. The implied probability might mismeasure true probability because the book is shading for public action. When both inputs are uncertain, combine them with other signals (model projections, streak data, matchup analysis) rather than treating either as authoritative.
6-step hit rate reading checklist
When you encounter a hit rate in a prop screener or tool, run through these six checks before acting on it.
1. What is the threshold? Is it measured against a fixed arbitrary number, or against the live sportsbook line? Line-relative hit rates are more actionable.
2. What is the sample size? 5 games: treat as a weak alert. 10 games: directional but uncertain. 20+ games: beginning to be meaningful. 50+ games: robustly informative.
3. What happened in the sample period? Were there lineup changes, injury absences, or a soft defensive schedule that might inflate the hit rate?
4. Is the cross-window pattern consistent? Does the hit rate look similar in last 5, last 10, and season? Consistency across windows is a much stronger signal than elevation in just one window.
5. What is tonight's implied probability? Does the hit rate exceed tonight's implied probability by enough to be interesting after accounting for sample uncertainty?
6. Is the situation tonight similar to the sample period? Is the player facing a similar defensive opponent, in a similar lineup context, on similar rest? A hit rate from a period when conditions were meaningfully different is less relevant tonight.
Hit rate reliability checklist
A hit rate is more reliable when it has all of the following properties:
✓ Measured against the live sportsbook line, not a fixed threshold
✓ Based on 20+ games (10+ is acceptable with caveat)
✓ Consistent across multiple time windows (last 5, last 10, season)
✓ Calculated from a sample period with a similar lineup context to tonight
✓ Not explained by schedule effects (soft defence opponents, fast-paced games)
✓ Supported by a structural reason (role change, usage increase) rather than pure variance
When a hit rate meets all six criteria, it is worth acting on. When it meets fewer, treat it as a hypothesis that requires additional research — not a bet.