Two honest ways to price a hockey game
The first way is to ask the market. In hockey the moneyline IS the market's probability statement: no spread, no handicap, just a price on each team to win the game outright, overtime and shootout included. Enter both teams' lines and this calculator converts each to an implied probability, adds them up, names the excess over 100 as the bookmaker's fee, and scales back to 100 for the fair, fee-free probabilities. Enter one line alone and you get the raw implied probability with an honest reminder that the fee is still inside it. (For converting between American, decimal, and fractional formats, or payouts on a stake, use our odds calculator; this page's job is the win probability.)
The second way is to model the game. Give each team an expected-goals figure for the night and we treat regulation as two independent Poisson goal counts, add up every scoreline, apply one disclosed correction for how often real games stay level, and send the tied games to overtime, where somebody always wins. You get each team's win probability, the chance the game reaches overtime, and the most likely regulation scorelines.
Why hockey does not get the football math
Our NFL and NBA pages take the point spread and divide it by a normal margin sigma, and for those sports that works: margins of victory are spread over a wide range and pile up in a roughly bell-shaped heap. Hockey breaks every assumption in that sentence. Margins are small integers produced by a low-scoring Poisson-like process; nearly half of NHL games are decided by a single goal. The "spread" (the puck line) is frozen at 1.5 by convention rather than moved to the expected margin, so it carries no information about how far apart the teams are. And since 2005 overtime abolishes ties, so there is no draw to distribute probability to. Fitting a normal curve to that is numerology, and a calculator that did it anyway would be dressing the wrong sport in football's clothes. Naming what a model refuses to do is as important as what it does, so this page runs the two methods above and nothing else.
The formula
From the moneyline, each price becomes an implied probability directly:
The two implied probabilities sum to more than 100 because the bookmaker's fee (the overround, or vig) is baked into both prices; dividing by the sum removes it.
From expected goals, each team's regulation goal count is Poisson with mean λ (its expected goals), and the scorelines multiply because the counts are treated as independent:
We sum scorelines to 30 goals a side, far past anything hockey produces, so the truncation error is negligible. One honest correction: independent Poisson counts underestimate how often real games are level after 60 minutes, because tied NHL teams play tight, careful hockey. So the tie share is scaled up by a factor of 1.5 and the three regulation outcomes rescaled to total 100. That single calibration makes an even 3.0 vs 3.0 game reach overtime 23.08% of the time (straight independence would say 16.67%), matching the roughly 23 percent of NHL games that actually get there. The overtime split by scoring rates is a disclosed simplification too: 3-on-3 hockey and the shootout scramble team strength toward a coin flip, so the split slightly flatters the favorite.
Worked example
From the moneyline: -180 vs +150. The favorite at -180 implies 180 ÷ 280 = 64.29%; the underdog at +150 implies 100 ÷ 250 = 40%. Together that is 104.29%, so the extra 4.29 points is the fee. Scale back to 100 and the fair probabilities are 61.64% for the favorite and 38.36% for the underdog.
From expected goals: 3.2 vs 2.6. Straight independence gives Team A ahead after regulation 51.39%, Team B ahead 32.12%, level 16.49%. After the tie correction: 47.47% / level 22.85% / 29.67%. The level games split 3.2 ÷ 5.8 = 55.17% to Team A, landing at Team A 60.08%, Team B 39.92%. The single most likely regulation scoreline is a 2-2 tie at 7.26%, which is the overtime story in miniature: the likeliest single outcome of a close hockey game is that regulation does not settle it.
What the puck line actually is
The puck line looks like hockey's point spread, and that resemblance misleads people every night. A football spread moves to wherever the market thinks the margin belongs: -3, -6.5, -13. The puck line is frozen at 1.5 goals for almost every game regardless of how lopsided the matchup is, and what moves instead is the price attached to it: a heavy favorite might be -1.5 at -110 while a slight favorite is -1.5 at +180. The information lives in the odds, not the handicap. That also means converting a -1.5 puck line price into a win probability is a different question from the one this page answers: it needs the probability of winning by two or more, which depends on the whole margin distribution, not just who wins. We would rather decline that conversion than fake it with a formula built for football; if you have the moneyline, use it, because it prices exactly the question "who wins the game."
Home ice, and what is already in the price
Home ice advantage is real and smaller than it feels: NHL home teams win around 53 to 55 percent of games, so being at home is worth only a few points of win probability, less than home field in football and much less than home court in the NBA. Resist the urge to add it on top of a moneyline result, because the market set that line knowing where the game is played; adding home ice again counts it twice. In expected-goals mode, home ice belongs inside your inputs: nudge the home side's expected goals up slightly and the visitor's down, and the model handles the rest. And the standing house reminder: everything on this page is probability math for fun, education, and settling arguments. The market's fee means the expected value of a bet is negative unless you genuinely know something the bookmaker does not, and a fair 61.64% favorite still loses nearly two games in five, which is why they play them.