Whet for the NHL.

A team's starter saves the puck at a ~3% better clip than the backup. Across 30+ shots a game, that compounds into ~1 goal of difference. Every prop on the board is priced against the assumption that the starter plays — if the backup starts unexpectedly, the prop math flips.

Whet won't surface an NHL prop until goalie confirmation hits (~30 minutes pre-puck-drop). When the backup is announced unexpectedly, the entire game's props re-score on the next scan (within ~5 minutes). The scanner respects the information.

A goalie on the second night of a B2B sees a ~2% drop in save percentage. Skaters drop ~7% in scoring chances. Whet's NHL filter applies a heavier penalty to goalies on B2B (-20% to their save-related props) than to skaters (-7% to scoring-related props).

Three games in four nights compounds. Late-season tired teams in playoff races get a heavier de-rate than mid-season fresh teams.

A team with a top-5 power play vs a bottom-10 penalty kill gets a meaningful boost on the high-PP-minutes skaters' props. The boost is calibrated on actual PP/PK rates for both teams over their last 10 games, not season averages (special teams move fast).

NHL shots-on-goal totals fluctuate ±25% per game. The per-game variance is high enough that small edges don't survive variance over fewer than ~50 bets. Whet's SOG filter requires 8%+ EV before surfacing (vs the 3% default), and flags the per-game shot total as a context input to the score.