NHL Model vs Market
NHL games where our model gives a team a meaningfully higher win probability than market odds imply. Showing disagreements above 5%.
No significant NHL probability gap yet
Our model shows a divergence when it gives a team meaningfully more probability than the market implies. Check back closer to game time.
How we find value bets
A value bet is a game where our model gives one team a higher win probability than the sportsbook's odds imply. Say our model makes the home team 60% to win, but the market price works out to 52%. That 8-point gap is the edge, and it's what we rank NHL games by. The bigger the gap, the more the model and the market disagree.
To compare the two fairly we strip the sportsbook's margin (the “vig”) out of the odds first, so we're measuring a true implied probability against the model's number, not a price padded in the book's favor. The implied probability explainer walks through that math with a worked example.
A gap is a starting point, not a lock
An edge means the model thinks the market is mispriced, not that the bet wins. A 60% team still loses 40% of the time, and our model is wrong plenty. We publish exactly how often on the accuracy page, down to which probability buckets we over- and under-shoot. Treat a divergence as one input you can check against your own read, not a tip to follow blindly. The methodology page shows the factors that go into every number.
Model outputs are for informational purposes only. Past performance does not guarantee future results.
See our model accuracy page for transparent performance data.