A month into 2025–26, teams with low pregame win probabilities are winning more often than the model expected. We looked at every game since October 1st from each team's perspective — both the home and away view of every matchup — and binned the results by pregame win probability.
The underdog uprising
| Pregame Win Prob | Games | Expected | Actual Win Rate | Performance vs. Model |
|---|---|---|---|---|
| 0–10% | 10 | 5.5% | 40.0% | +632% |
| 10–20% | 22 | 16.9% | 45.5% | +169% |
| 20–30% | 13 | 25.4% | 61.5% | +142% |
| 30–40% | 13 | 36.2% | 53.8% | +49% |
The headline: teams given less than a 40% chance to win are winning exactly half the time — 50% across 58 underdog situations, against a 21.2% average expectation. The longest shots tell the same story. Ten times a team entered at 10% odds or worse; four of them won.
What's driving it
Early-season hockey is noisy. Rosters are still settling, new systems aren't fully installed, and offseason rust lingers. With small samples, one hot goalie or a fluky bounce moves a whole probability bin.
The "Performance vs. Model" column shows the relative difference between observed and expected win rates. It describes this sample only. With just 10 to 22 team-game observations per bin, a few outcomes can move the percentages sharply; these results do not establish that the pattern will persist.
The bottom line
This covers games since October 1st, both home and away perspectives. The early returns look like the Year of the Underdog; the likelier outcome is that things settle as teams find their level. Hockey is random, but not this random. Still — if your team is a heavy underdog tonight, the first month says don't count them out.
Get NHL forecasts in your inbox
Occasional emails with NHL forecasts, analysis, and model updates.