We read 4 NHL preseason games: losers out-hit winners every time — and the winning goalie had the higher save rate in all 4

We pulled the ESPN NHL scoreboard at capture time on September 24, 2026 and read every game listed as Final — 4 games played September 23–24, all in the 2026–27 NHL preseason. These 4 are the full set ESPN's scoreboard listed as Final at that capture — nothing excluded, but a one-evening snapshot, not a population. Results: Toronto Maple Leafs 4, Ottawa Senators 2 (Canadian Tire Centre, Sept 23); Toronto Maple Leafs 6, Ottawa Senators 1 (Scotiabank Arena, Sept 23); Dallas Stars 2, Minnesota Wild 0 (American Airlines Center, Sept 24); Los Angeles Kings 2, Anaheim Ducks 1 (Honda Center, Sept 24). In each of the 4 games, the losing team registered more hits than the winning team: Toronto at Ottawa — Toronto won 4–2. Toronto recorded 11 hits; Ottawa recorded 30. Ottawa out-hit Toronto by 19. Ottawa at Toronto — Toronto won 6–1. Toronto recorded 23 hits; Ottawa recorded 26. Ottawa out-hit Toronto by 3. Minnesota at Dallas — Dallas won 2–0. Dallas recorded 16 hits; Minnesota recorded 17. Minnesota out-hit Dallas by 1. Los Angeles at Anaheim — Los Angeles won 2–1. Los Angeles recorded 19 hits; Anaheim recorded 32. Anaheim out-hit Los Angeles by 13. Across the 4 games, winners registered 69 hits combined; losers registered 105 — 52% more. In the same 4 games, the winning team's goaltender had a higher save percentage than the losing team's goaltender in every game: Toronto at Ottawa — winning goalie .905 vs losing goalie .870. Ottawa at Toronto — winning goalie .929 vs losing goalie .833. Minnesota at Dallas — winning goalie 1.000 vs losing goalie .917. Los Angeles at Anaheim — winning goalie .960 vs losing goalie .895. Shot efficiency (goals divided by shots on net): winners converted at 16.7%, 16.7%, 8.3%, and 10.5% in the four games; losers converted at 9.5%, 7.1%, 0.0%, and 4.0%. In every game, the winner converted a higher share of its shots. Power play across the slate: 3 goals on 32 opportunities, a 9.4% rate. Six of the 8 team-game units scored 0 power-play goals. Two winners — Dallas and Los Angeles — won without scoring a power-play goal.

The following is analysis, not fact. Read across these 4 games, hits — the physical category often taken as a proxy for winning the puck ^sspeculative — ran against the scoreboard in every instance. Goaltending, measured by save percentage, ran with the scoreboard in every instance. The pattern is clear in the two Toronto–Ottawa games. Ottawa landed 30 hits in the first game and 26 in the second; Toronto landed 11 and 23. Ottawa lost both. The Anaheim–Los Angeles split is also sharp: Anaheim landed 32 hits to Los Angeles's 19, outshot Los Angeles 25 to 19, and still lost: it converted only 1 of those 25 shots (4.0%) against a goalie stopping 96.0% of what he faced. In the Dallas–Minnesota game, goaltending appears to be the differentiator ^mmodeled. Both teams put 24 shots on net — an exact tie. Minnesota won 64.4% of faceoffs, suggesting territorial control ^sspeculative. Dallas gave the puck away 27 times against Minnesota's 13. On paper, these inputs might be read as favoring Minnesota ^sspeculative. Dallas won 2–0 with its goaltender stopping all 24 Minnesota shots while Minnesota's stopped 22 of 24 Dallas attempts. The 9.4% power-play rate is well below the NHL regular-season average of roughly 20% ^mmodeled, consistent with early preseason patterns where first-unit personnel often share ice time with depth forwards and the unit's timing is still being established ^sspeculative. It need not indicate a structural problem ^sspeculative. None of these 4 games are a meaningful sample for regular-season projection. Early preseason rosters include tryout players and developmental personnel whose regular-season roles are unsettled. A goaltender who appears in one preseason game under low-stakes conditions is not necessarily the starter who takes the ice in October ^sspeculative.

Hits are a volume stat, not a quality stat — a body check that separates a player from the puck counts the same as one that does not. A team accumulating more hits may be chasing play rather than dictating it, in which case higher hit totals are consistent with being behind, not a cause of losing ^sspeculative. The hit–outcome relationship in these 4 games is a correlation, not a mechanism, and 4 games is too small a sample to conclude anything about hits as a predictor across a season. Preseason goaltender numbers are also noisy. A goalie stopping .929 in a low-pressure preseason game, sometimes facing a second or third offensive line, is not the same as a playoff goalie at .929. The save-percentage pattern here is real in this box score, but its predictive weight for the regular season is low ^sspeculative. A reader could fairly point to the Ottawa–Toronto series as a two-game matchup where the same physical imbalance produced the same scoreboard outcome — which is more suggestive than a cross-team sample of 4 — but it is still 2 preseason contests between the same clubs, where familiarity effects and lineup experimentation both apply ^sspeculative.