We use hockey statistics and computer-based predictions to estimate each team’s chance to win. Then we show you how those predictions actually perform.
See each team’s chance to win and compare our three approaches.
BUFBuffalo Sabres
OTTOttawa Senators| Team | Moneyline | Puck line |
|---|---|---|
| Buffalo Sabres | -104 | — |
| Ottawa Senators | +104 | — |
A 57% chance does not mean a guaranteed winner.
FLAFlorida Panthers
CHIChicago Blackhawks| Team | Moneyline | Puck line |
|---|---|---|
| Florida Panthers | -208 | — |
| Chicago Blackhawks | +208 | — |
A 57% chance does not mean a guaranteed winner.
LAKLos Angeles Kings
SJSSan Jose Sharks| Team | Moneyline | Puck line |
|---|---|---|
| Los Angeles Kings | -117 | — |
| San Jose Sharks | +117 | — |
A 57% chance does not mean a guaranteed winner.
SEASeattle Kraken
CGYCalgary Flames| Team | Moneyline | Puck line |
|---|---|---|
| Seattle Kraken | -208 | — |
| Calgary Flames | +208 | — |
A 57% chance does not mean a guaranteed winner.
Times shown in Eastern Time. These are reporting categories, not claims that any prediction is safe or guaranteed.
Real results from predictions recorded before games began.
Correct winners among finished, archived predictions
Brier score: lower is better. 0 is perfect; 0.25 is the score of always predicting 50–50.
Only eligible, pregame predictions are published in our track record.
| What the predictions consider | Finished predictions | Accuracy | Brier score |
|---|---|---|---|
| Recent Performance | 80 | 52.5% | 0.2618 |
| Overall Team Performance | 91 | 50.5% | 0.2485 |
| Power Plays & Penalty Killing | 52 | 50.0% | 0.2482 |
Three approaches in our AI-assisted hockey forecasting project. Each uses a different set of information. Here is exactly what each considers.
Rules-based calculation • part of our AI-assisted project
Our original, hand-built statistical formula. It applies the same rules to every matchup.
Trained prediction system • machine learning
A prediction system trained on past NHL games to combine several team-performance factors.
Trained prediction system • machine learning
A separately trained version of Overall Team Performance that also considers past power-play scoring and goals allowed.
Starting goalies, injuries, and last-minute lineup changes may not be included.
Our Recent Performance formula is rules-based, not a trained AI model. Overall Team Performance and its special-teams version use statistical machine learning trained on past results; they do not use a chatbot to make the picks. Special-teams inputs count power-play goals per game, not official power-play or penalty-kill percentages. Each approach estimates a chance of winning, not a guaranteed result.
Every prediction has a probability. Every probability has a track record. We publish the good results and the bad.