Across the three-day window from October 11 to October 13, 2026, the nhl_selective_v1 model posted a 6-9-0 record across 15 spread picks. Despite a 40.0 percent win rate, the model generated a net profit of +3.62 units. A sample of 15 games across three days is too small to validate or invalidate a model, but auditing the individual results shows how pricing mechanics dictated the overall return.
High-Odds Hits Build Early Capital
The positive unit return was driven entirely by plus-money spread covers. On October 12, Buffalo (-1.5) hit at +233 odds in a 5-3 victory over Florida, delivering +2.33 units. Tage Thompson recorded a goal and an assist, while Jack Quinn and Owen Power provided two assists each for Buffalo. Sam Reinhart scored two goals and Seth Jones added two assists for Florida in the loss. On the same night, Minnesota (-1.5) converted at +245 odds in a 4-2 win over Vegas behind two assists from Quinn Hughes, adding +2.45 units.
The model's third October 12 selection, New Jersey (-1.5 at +220), resulted in a 1.00 unit loss following a 4-1 defeat to Ottawa, where Tim Stützle posted a goal and an assist.
On October 11, the model split its two selections for a net gain of +0.33 units. Carolina (+1.5 at +133) covered in a 4-1 win over Philadelphia, backed by Andrei Svechnikov (1 goal, 1 assist) and William Carrier (2 assists). Washington (-1.5 at +133) failed to cover in a 5-3 loss to Seattle, where Frédérick Gaudreau and Ryan Winterton each scored a goal and an assist, and Ben Meyers added two assists.
Navigating the October 13 Slate
The October 13 slate accounted for 10 of the window's 15 picks, yielding three wins and seven losses for a net loss of 0.49 units. Final scorelines were not recorded in the tracking system for these games, but actual margins and line gradings reflect the settled outcomes cleanly.
Three plus-money underdog covers accounted for all positive units on the day: Edmonton (+1.5 at +208) covered with an actual margin of -3 (+2.08 units), St. Louis (+1.5 at +228) covered with an actual margin of -5 (+2.28 units), and Boston (+1.5 at +215) covered with an actual margin of -3 (+2.15 units).
The slate's highest confidence pick was Washington (-1.5 at -170) with a confidence score of 0.63, which failed to cover with an actual margin of 2, losing 1.00 unit. Six other spread attempts lost 1.00 unit each: Dallas (-1.5 at +239, margin -1), Detroit (-1.5 at +239, margin -1), Montréal (-1.5 at +174, margin 1), Tampa Bay (+1.5 at +170, margin 3), Philadelphia (-1.5 at +233, margin -1), and Minnesota (+1.5 at +233, margin 3).
Window Summary
| Market | Picks | Record | Win % | Units |
|---|---|---|---|---|
| Spread | 15 | 6-9-0 | 40.0% | +3.62 |
While a return of +3.62 units on 15 picks is positive, a 40.0 percent hit rate over three days is simply short-term noise. The NHL model remains under active audit, and long-term conclusions require a significantly larger sample size.
More from the Sharp Report: all NHL articles