NHL Referees Tonight

For power plays, a little: The full test (one season against the next) runs once every season's crews are loaded. On the 1,747 games loaded so far, true referee differences look small (about ±0.22 power plays per game for both teams combined) and only weakly repeatable (odd vs even games: r = 0.18 across 39 referees). Until it passes, the effects are descriptive only and are not used in the power-play forecast.

For totals, no. Over 2,618 games with closing lines, the 38 referees with 60+ games differed from each other by no more than chance would produce (raw spread SD 0.16 goals per game; estimated true spread 0.00). One season's goals against the line predicted the next at r = −0.19, and over rates at r = 0.04 (34 referees).

Because no referee effect on goals survives the repeat check, every referee's shrunk totals effect is zero and there is nothing to bet against the closing total. The O-U-P records above are what happened, with a noise band that crosses zero for every referee. For the totals market itself, blind overs and unders both lost after the vig: over/under records.

Power-play props: the expected power plays and PP1 minutes for every team tonight are on Power Play Tonight. More: NHL betting research.

For each referee working tonight: power-play opportunities and penalties per game in their games, the shrunk effect on power plays against what the two teams would normally draw, and goals in their games against what the closing totals expected. Crews are posted close to puck drop, so the page fills in through the evening.

Power-play effects are small and shrunk hard for sample size; the section below says how well they repeat. Totals are another matter: over 2,618 games with closing lines, the spread between referees was no wider than chance, and one season predicted the next at r = −0.19.

A whistle-heavy crew can nudge power-play props and PP points, a little. It is not a reason to bet an over: books do not shade totals for referees, and the data says they do not need to.