Free toolshalf-point-value
Books sell half-points at a fixed price. Games do not land on every number equally: an NFL game finishes exactly three points apart about once every six or seven games, and two points apart about once every twenty. This tool reads our measured record of every completed NFL and NBA game, shows how often the number you are crossing has actually come up — with its sample size and its 95% confidence interval — and works out what that frequency prices the move at. Then you compare it to what you are being charged. It is arithmetic on frequencies, not a prediction about a game.
League
Market
A half-point move always crosses exactly one whole number. 3 is that number.
−3 to −2.5: a 3-point win used to push, now it wins. The underdog side is the mirror image — the arithmetic is identical, which is why this control asks what happens to the number rather than asking for two signed lines.
Everything below is anchored on this price. It prices the half-point, not the bet — it cannot know whether the number you already have is itself a fair one.
What you would actually pay after the half-point move.
Loading the measured distribution…
Leave blank to price off the measured history above. Fill it in to price off a number you trust more — the panel below always says which one it used.
Enter a whole number with a published frequency (or type your own), plus a non-zero American price at your current line, to price the half-point. Some combinations have no answer: a half-point cannot move more probability out of a bucket than the bucket holds.
These are frequencies across all games, not conditioned on the closing number. Buying a half-point off a line that was never near 3 does not buy you 15.4%. We publish it unconditionally on purpose: our stored closing spread is an average across books, so “games priced at exactly −3” is not a set this data can identify, and rounding that average to fake the condition would be a manufactured statistic rather than a measured one.
NFL and NBA behave completely differently, and the NBA answer is the duller one. NFL scoring is built out of 3s and 7s, so a handful of margins carry several times the weight of their neighbours. NBA margins are close to flat: the most common margin is barely more common than an ordinary one, which means an NBA half-point is worth roughly the same wherever you buy it. Toggle the league above and read the two histograms next to each other — that contrast is the measurement, not a framing.
A bet that returns your stake on a push has three outcomes, not two. Write them as win, push and loss. The useful fact is that an American price's implied probability is exactly the win probability conditional on not pushing — because a push returns the stake and simply drops out of the arithmetic:
So the price you already have, plus the measured frequency of the number, pins the whole three-way split. A half-point then moves exactly that frequency of probability one step — a push becomes a win, or a loss becomes a push — and the fair price of the new split follows from the same identity:
Two consequences are worth naming. First, everything is anchored on the price you start from: the tool prices the half-point, not the bet, and it has no opinion on whether your starting number was a good one. Second, buying onto a number (a loss becomes a push) is worth slightly more than buying off it (a push becomes a win), because getting your stake back is worth a full unit while winning is worth only the profit multiple — and on a −110 favourite that multiple is less than one.
NFL scoring is assembled out of 3s and 7s, and the margins inherit that structure: three is far and away the most common final margin, seven is the next spike, and the numbers in between are ordinary. That is why a half-point across 3 costs real money and a half-point across 5 costs almost nothing — and why a book charging one flat price for both is charging the same for two very different things.
NBA margins do not work like that at all. Across our sample the most common margin is barely more common than a middling one, so the distribution is close to flat and every half-point is worth roughly the same. That is a dull sentence, and it is also the honest one: there are no NBA key numbers to hunt for. Most tools in this category quietly present the NFL story with a league toggle bolted on. The histogram here changes shape when you flip the toggle because the underlying measurement does.
The frequencies are unconditional: they count how often every completed game landed on each number, across the whole window. They are not conditioned on where the game was priced. We would like to publish the conditional version — “how often did games priced at exactly −3 land on 3” — but our stored closing spread is an average across books, which produces numbers like −8.1 that no book ever posted. There is no honest way to recover “games priced at exactly −3” from that, and rounding the average to fake it would invent a statistic. So the unconditional number is what we publish, and the caveat travels with it on the page rather than living in a footnote.
Coverage is NFL and NBA only. Those are the two leagues whose final scores are complete end-to-end in our data; every other league we ingest is missing between 10% and 100% of its finals, and a distribution built on a file like that would be a guess wearing a percentage sign. Sample sizes and date windows are printed next to every number for the same reason — a rate without its denominator is a marketing figure. If you want the fair-price side of the same question, the no-vig calculator strips the margin out of a two-way price, and the methodology page explains how we grade everything we publish.
More free tools
These calculators run the exact helpers the Blitzen terminal uses on every board. See them applied to real markets — and how we grade everything against the close.