Every NHL game Kalshi lists, priced off one model — Goal Light — and shown three ways: the market, the model, and the market-anchored fair value. Starting goalies are checked before any number publishes.
Quick Answer
2 NHL games tonight; the model prices the slate at 10:45 AM ET once the projected starters post.
Prices are Kalshi top-of-book mids in cents. Fair = market-anchored model. A position needs 2pp net of the spread; past 7pp we make no call. Taker fee 7% × price × (1 − price) per contract, max 1.75¢. About Kalshi.
Not priced yet — the board fills at 10:45 AM ET on game day.
Where it trades — New York Rangers to win: Kalshi 70.5¢ (ask + fee) · Novig 68.0¢ (no fee pregame) · Novig 2.5¢ cheaper (as of 6:30 AM ET) · Novig → · NHL on Novig
Not priced yet — the board fills at 10:45 AM ET on game day.
Where it trades — Vegas Golden Knights to win: Kalshi 63.7¢ (ask + fee) · Novig 62.0¢ (no fee pregame) · Novig 1.7¢ cheaper (as of 6:30 AM ET) · Novig → · NHL on Novig
Kalshi has not listed tonight's player-goal contracts yet. The board fills when they open.
Every goal-scorer contract →Two records, graded apart: game contracts and player goals. Gross P/L on one contract per position; wins and losses both.
Game contracts
Sample size: building · 26/100
Goal of the Day
153–129 · net +$2.05 · avg +0.7¢ per contract · CLV -0.8pp
Full leaderboard, every engine: /leaderboard/tools.
Size a position at ¼-Kelly on the fair probability.
Expected value of a contract at the Kalshi ask.
Kalshi vs Polymarket NHL moneyline gaps.
The two venues on the same games.
Tonight’s games priced on Kalshi and Novig, and which is cheaper.
The same board — market, model and fair value, goalies checked — answers inside Claude, ChatGPT, Grok or Cursor through the free PredictionMarketsPicks MCP server: one connector, no API key. Ask for tonight’s games, the goal contracts or the Stanley Cup futures.
The NHL MCP server →Goal Light rates every club on expected goals, not goals: a weighted Poisson regression on regulation, non-empty-net xG with a 45-day half-life on a hockey calendar, home ice and rest fitted rather than assumed. Finishing is shrunk hard toward the league. The starting goalie is his own layer — goals allowed against expected, on every unblocked attempt he faced.
One score grid comes out of those rates, with the empty-net map and the overtime split measured on 2,624 games, and every contract on the board is read off that grid — so the moneyline, the puck line, the totals and the overtime market cannot contradict each other. Walk-forward on the 2025-26 season the moneyline graded 0.681 log-loss against a 0.692 base rate; win, puck line, team totals and player goals carry skill, the rest publish as references.
Kalshi charges a trading fee on each fill, not on winnings: 7% × contracts × price × (1 − price), rounded up to the cent, so it peaks at 1.75¢ per contract on a 50¢ contract and shrinks toward zero at the extremes. It is paid at entry whether the position wins or loses; there is no settlement fee. Resting (maker) orders pay no fee on most standard markets. Educational analysis, not financial advice. Trade responsibly.
The moneyline includes overtime and the shootout, exactly as the Kalshi contract settles. The grid prices regulation first, then 68% of ties go to overtime (the better 3-on-3 team wins more often) and 32% to a shootout priced as a coin flip.
The NHL credits the shootout winner one goal in the final score, and Kalshi settles totals on the final score. So an over 5.5 contract on a 3–2 game decided in a shootout settles yes at 6.
Fair is the market-anchored value: the books' de-vigged consensus (or the Kalshi mid when fewer than three books quote) moved part of the way toward the model. A position needs the fair value to clear the Kalshi price by 2 points net of the spread; past 7 points we make no call, because a disagreement that large with a liquid price is usually ours.
Game totals, overtime, first-period and first-ten-minute markets price close to the league base rate in our backtest — the model cannot tell one game from another on them yet. We publish the number as a reference and grade it; a family earns positions only when the graded evidence says so.
On a full season of starts the fitted goalie factors run from about 0.92 to 1.06. Swapping a top starter for his backup moves a league-average game by up to 4–5 points on the moneyline and a quarter of a goal on the total, which is why the board re-prices four times a day as starters get confirmed.
Kalshi lists more NHL contracts than Goal Light prices today — 2nd- and 3rd-period winners, period totals and spreads, and player points, assists and goalie saves. We show every one of them at the live Kalshi price so the whole board is in one place, labelled "market price, no model yet". No call is ever made on a row the model has not priced.
A fork means a starting goalie is unconfirmed, so we publish both lines — with the projected starter and with the other goalie — instead of guessing. Positions are withheld on a forked game until the starter is confirmed; the leans stay visible.
Every published position is graded against the Kalshi contract that priced it, under two records: game contracts (NHL Goal Light) and player goals (NHL Goal of the Day). Each record goes public once it reaches 100 graded positions.
No. PredictionMarketsPicks is the analysis layer. The contracts trade on Kalshi, a CFTC-regulated exchange; every row links to its Kalshi market. Trade responsibly — an edge is disagreement with the market, not a guarantee.
Prices refresh through the day; the model re-prices at 10:45 AM, 4:15 PM, 6:35 PM and 9:35 PM ET as starters are confirmed. First puck drop tonight 9:00 PM ET.
