Wed, Oct 8, 2025
Archived sheet
1 of 4 games had posted player lines
Archived Shots on goal
4 games · 10 players · 6 posted SOG lines. These are the lines as they stood on that date — not current, and not a forecast for any game being played now.
The model has not beaten the Pinnacle closing line: shots-on-goal ROI −0.0709 [−0.0932, −0.0467] per unit over 7,171 bets in a 2025-26 backtest of 512 games. How this was measured
4 games · 10 players · 6 posted SOG lines · sorted by projection
Line is the book's posted threshold · Proj is the model's mean · P(over) is the modelled probability of finishing above the line. Each bar is P(SOG = k) — the whole distribution, not a pick. Research context, not a recommendation.
Montréal Canadiens at Toronto Maple Leafs
7:00 PM ETGame pageNo props in this market for this game.
Boston Bruins at Washington Capitals
7:30 PM ETGame pageNo props in this market for this game.
Calgary Flames at Edmonton Oilers
10:00 PM ETGame pageNo props in this market for this game.
Los Angeles Kings at Vegas Golden Knights
10:00 PM ETGame page
Line
2.5
Proj
2.7
P(over)
50%
Line
2.5
Proj
2.4
P(over)
43%
Line
1.5
Proj
2.1
P(over)
60%
Line
1.5
Proj
2.0
P(over)
57%
Line
1.5
Proj
1.8
P(over)
51%
Line
1.5
Proj
1.5
P(over)
42%
No line posted.
No line posted.
No line posted.
No line posted.
NHL · Shots on goal
Distributions, not picks.
OddPuck projects a full shots-on-goal distribution for every skater on the slate — P(SOG = k) against the book's posted line — from five seasons of play-by-play. No units, no plays of the day. You read the shape and draw your own conclusions.
They were measured against the Pinnacle closing line, which they do not beat — shots-on-goal ROI −0.0709 [−0.0932, −0.0467] per unit over 7,171 bets in a 2025-26 backtest of 512 games, and the model's disagreement with the market carries no signal a paired test can separate from betting the other side instead. That is the measured result, reported rather than buried.
The signature — P(SOG = k) vs the line
Bars right of the line clear it; each height is the modelled probability. Positional, never a verdict.