The Empty-Net Factor: How Final Minutes Change Hockey Odds
A goalie pull changes the players on the ice and leaves a net unguarded. An ensuing goal can change both the final margin and the total. Understanding those possible score paths is useful without assuming a league-wide empty-net rate or an automatic pricing advantage.
Separate possible paths from their frequencies
Replacing a goaltender with an extra skater is the mechanism discussed in the NHL’s coaching account of goalie-pull decisions. That account is qualitative coaching context, not a dataset for the numerical rates removed from this guide.
Source: Paul MacLean’s NHL coaching account, 2018.
Hypothetical paths from a 3–2 lead: no further goal leaves margin one and total five; a goal by the leader leaves 4–2, margin two and total six; an equalizer leaves 3–3 before subsequent play. These are possible score transitions, not an exhaustive terminal-outcome model or assumed probabilities.
The 4–2 path covers the leader at -1.5 and goes over 5.5, while the 3–2 path does neither. That arithmetic explains why the event can matter. It does not tell you how likely either path is at the price offered.
Define denominators before measuring a rate
A study should distinguish games with at least one empty-net goal, number of empty-net goals, goalie-pull episodes, and time spent with a net empty. These have different denominators. State seasons, regular-season or playoff scope, score deficit, time remaining, and whether delayed-penalty situations are included.
A conditional rate given a pull is not the unconditional pregame probability. Forecasting the latter also requires the probability of reaching the relevant state and of a pull occurring there. Repeated pulls and goals in one game need explicit treatment.
Subtracting empty-net goals from recorded final scores is a descriptive score adjustment. It does not recreate games in which no goalie was pulled: possession, tactics, subsequent goals, and elapsed time could all have changed. Do not label that subtraction a causal lift in cover probability.
Connect the forecast to a contract
A usable forecast needs the distribution of final scores under the relevant overtime and shootout treatment, not just an expected number of empty-net goals. Compare the resulting win, push, and loss probabilities with the quoted payout. An empty-net possibility can already be reflected in that price.
This guide establishes no universal pull time, equalizing frequency, home/away cover difference, or structural edge. Such claims need event-level data and a defined estimation method. It also does not promise live wagering on OwnTheLines: its simulated selections are subject to the product’s eligibility and start-time rules.
Continue reading: The Puck Line Guide · The Logic of Line Movement.
Frequently Asked Questions
Does removing recorded empty-net goals show what would have happened without a pull?
No. It changes the recorded score but does not model how play would have developed under a different decision.
Is a rate conditional on a goalie pull a pregame probability?
No. A pregame forecast must also account for reaching a pull situation and the decision to pull, using clearly defined states.
Does an empty-net forecast establish a structural edge?
No. Value still depends on a complete outcome distribution, the actual payout, settlement rules, and uncertainty.