Every skater on the slate with the model's chance that he scores at least once tonight. An XGBoost classifier reads his recent goals, shots and ice time, his power-play scoring and the opponent's goals and shots allowed; the result is a calibrated probability, the number to hold against your book's price.
The percentage is the model's own chance that he scores at least once, not a sportsbook price — compare it with the odds your book is offering. Graded picks are settled against the box score after the night's last game is final, on the performance page.
Every skater on the slate with the model's chance to score, and its graded picks, updated three times a day. See plans, or read how the projections are built below.
See plans →The model is a classifier fit on about two and a half seasons of skater games. Its inputs are the player's goals over his last five and ten games and this season, how often he has scored at all, his shots, points and hits over his last ten, his ice time recent and season-long, his power-play goals, games played, home or away, rest and back-to-backs, his position, and the opponent's goals and shots allowed over their last ten games. A skater needs ten games played before he gets a number.
The probability is calibrated: across 9,689 archived player-games, skaters the model puts at 15 to 20% scored 19.8% of the time and those it puts at 30 to 35% scored 34.1%. That matters more than it sounds. In a market where nearly every price is plus money, a number that runs a few points hot turns every pick into a bad bet.
A skater scores in about 15 games out of every 100. Forwards score in 19%, defensemen in 7%. Even the league's best finishers — players scoring half a goal a game or more — found the net in only 39.2% of their games over the last three seasons. So a 35% chance on this board is a strong number, not a weak one.
That is also why the price is the whole bet. A 30% chance breaks even at +233; a 35% chance at +186; a 40% chance at +150. A top scorer at +150 is priced as a 40% chance, before the book's margin. The full base rates by position, role, shot volume and scoring rate are in our anytime goal scorer guide.
Each card shows the model's chance he scores tonight and his last ten games, with a green bar for every game he scored in. A "Graded play" tag marks a skater at 30% or better. Top-six forwards scored in about one game in four over the last three seasons, so two or three green bars in ten is normal for one; the model weighs a short hot or cold run against his longer scoring rate.
The graded anytime goal play is every skater the model gives a 30% or better chance to score, with no daily cap. Each one is settled against the official box score after the night's last game: one or more goals wins, none loses. A skater who is scratched or never takes a shift is no result, not a loss. Shootout goals do not count; empty-net goals do.
The rule started on October 15, 2026. Replayed over the archived predictions before that date, it scored 224 times in 640 picks (35.0%). Expect a record in the mid-30s to around 40% — that is what a good anytime goal record looks like. The live record sits on the performance page with every other market.
About 15% of the time. Over three regular seasons and 141,675 skater games, a skater scored at least once in 15.1% of them. Forwards scored in 19.2% of their games and defensemen in 7.0%. Teams averaged about three goals a game, spread over 2.7 different scorers.
Anything above 30% is high. Players who had scored half a goal a game or more over their previous 82 games found the net in 39.2% of their next games. Our graded rule takes every skater at 30% or better, and its replay over the archive hit 35.0%.
Every skater the model gives a 30% or better chance to score is a graded play. It wins if he scores at least once in regulation or overtime, settled against the official box score; shootout goals do not count. A skater who is scratched or never takes a shift is no result.
Convert the chance to a break-even price. A 30% chance breaks even at +233, 35% at +186 and 40% at +150. If the book's price is shorter than the break-even for your probability, the bet loses money over time even when the player is a good bet to score.
Rarely. Defensemen scored in 7.0% of their games over three seasons, against 19.2% for forwards. Their prices are long for that reason. A defenseman who shoots a lot and plays the first power-play unit can sit well above that rate, which is what the model reads.
Shots on goal, goalie saves, goals, assists, points and hits, plus live in-game signals. MLB, NFL and NBA included.
See plansFor information only. 21+. If gambling is a problem for you or someone you know, call 1-800-GAMBLER.