This page is loading — it needs JavaScript. The written page below is served without it.
A hit rate is one fraction: how many games a player finished above a stated number, over how many games were counted. Both halves have to be on screen for it to mean anything — the number it was measured against, and the games it was measured over.
Take a player’s game log, pick a threshold, and count how many games finished above it. That count over the number of games counted is the hit rate. Nothing about it is a projection: it is a summary of games that have already been played, and it stays true for as long as those games stay played.
Which means a hit rate on its own is not a fact yet. "Cleared it seven times" is not a measurement until you know out of how many, and "cleared it in 70% of games" is not one until you know the threshold it cleared. A number quoted without both halves is a number that has been chosen for you.
A three-game denominator admits only four possible answers: 0%, 33%, 67% and 100%. A player who happened to clear a line three times in a fortnight reads as a perfect record, and a player who missed once in three reads as a slump. Neither reading survives a fourth game, and the shorter the window the more of them look extreme.
This is the single most common way a prop number misleads, and it does not require anybody to be dishonest — a last-three column is genuinely useful and genuinely unstable at the same time. The defence is to read the denominator first and the percentage second, which is why every rate published on this site is printed with the number of games beside it, and why the pages lead with a deep window rather than the flattering one.
A threshold can be chosen two ways: from the number a sportsbook has posted, or from the player’s own distribution. They answer different questions. The posted number asks "how has he done against what is on offer right now", which is the question a bettor eventually needs — and it is also a moving target, because the line changes through the day and a page written this morning would be describing a number that has since gone.
The player’s own distribution asks "what does this player actually do", and it has neither problem. The measured pages here build a ladder of thresholds from the game log and lead with the rung nearest a coin flip rather than the rung with the highest rate. That is deliberate: a page that opens with whichever threshold flatters the player has stopped describing him. The high rung always exists, because any ladder long enough contains one.
So the two numbers do different jobs and both are on the site. The measured history is baked into the page and rebuilt daily from box scores; anything about a posted price is loaded live, because it is only true while it is on offer.
A last-five and a last-fifteen answer different questions, and a gap between them is information rather than noise to be averaged away — it says the recent games differ from the longer record, which is a starting point for asking why rather than a conclusion. Minutes, role, rest and opponent all sit behind that gap and none of them is visible in the rate itself.
Splits inherit the same denominator problem and make it worse, because every split divides the sample. A home-and-away split of a fifteen-game log is two seven-game samples; a split by opponent is often two or three games. They are worth reading in the direction of "is there anything here at all", and not in the direction of a precise number. The splits on this site print their own game counts for that reason.
It is not the probability of tonight. It is a count of what happened in a set of games that are over, under conditions that will not repeat exactly, and the player’s role may have changed inside the very window being counted. Treating a rate as a forecast is the mistake the number makes easiest.
It is also only half of any comparison. The other half is the price, and a price is a percentage in disguise — the implied probability it carries, margin included. A rate without a price says nothing about whether a bet is worth taking, and a price without a rate says nothing about the player. Parlay Builder publishes both and does not accept wagers, hold funds or settle bets; past performance does not predict future results.
The share of counted games in which a player finished above a stated threshold — games cleared over games played. It is a summary of completed games rather than a forecast, and it only means something when the threshold and the number of games are quoted with it.
That question has no answer without a price, and this site will not invent one. A high rate against a short price and a lower rate against a long one can be the same bet; a rate on its own tells you about the player and nothing about the offer. Anyone quoting a threshold you should require has stopped describing the number and started recommending a bet.
Because a three-game denominator admits only four possible answers, one of which is 100%, so a short window produces extreme readings for ordinary players. The pages here lead with a deep window and print the game count beside every rate, which makes the numbers less flattering and considerably more stable.
The measured history on the page is not — it uses thresholds built from the player’s own game log, so it stays true after the line moves. Anything measured against a posted line is loaded live from the board instead, because a page cached overnight cannot honestly name a number that changed at lunchtime.
Official box scores, one row per game a player actually played, loaded daily and counted directly. No projection, no model output and no third-party rating enters the count, which is what makes a rate here checkable against the game log printed underneath it.
Every prop the books have posted, each with the measured history and the game count behind it. See tonight’s board.
You must be 21+ (or older, where your jurisdiction sets a higher age) to bet. If gambling stops being fun, call or text 1-800-GAMBLER for free, confidential help. All guides.
Odds and lines come from licensed data feeds and refresh on a schedule; hit rates, edges and grades are Parlay Builder’s own calculations.