New York Islanders head coach Patrick Roy has never tried to sell himself as an analytics-driven coach. If anything, he’s often portrayed as the opposite — an instinct-first competitor who trusts his eyes, his bench, and his feel for the game.
But after Tuesday’s 4-2 loss to Calgary, Roy showed that while he may not worship advanced metrics, he understands them well enough to challenge their shortcomings.
The moment in question came when Roy opted for an aggressive decision in the third period, pulling his goalie with over 8 minutes left and the team trailing by three goals. Asked whether that move was driven by analytics or instinct, Roy was direct. “Sometimes you just go with your gut feeling and then try something different,” Roy said. “That’s all. Nothing really personal there.” Pressed further on whether pulling the goalie early was influenced by data models or probability charts, Roy didn’t flinch. “I believe that we’re going to win the game,” he said.
That belief, more than any spreadsheet, guided the decision. But what followed was the most revealing part of Roy’s postgame comments. For a coach who is often framed as dismissive of analytics, Roy spoke fluently — and critically — about one of the sport’s most cited metrics.
“And for analytics, I’ll give you an example,” Roy said. “Tonight, you’re looking expected goals against. The first one — there’s a tip they gave 20-something that never hit the net. I mean the wraparound they gave 17 that never hit the net.”
Roy’s point wasn’t that analytics are useless. It was that they are incomplete without context. Expected goals can suggest danger, but they don’t always reflect reality. "I don't give a sh*t about analytics, to be honest," he later added, though he sure did know what they were on Saturday.
Shots that miss the net entirely don’t force saves. Chances that look threatening on a chart don’t always translate to real scoring opportunities. Per Natural Stat Trick, the Isles had a 13-4 edge in high-danger chances skating five-on-five and 18-8 in all situations. It's a game they deserved to win according to the advanced analytics, just almost as much as they deserved to lose versus Edmonton.
For Roy, coaching remains a balance between information and instinct. Analytics can inform decisions, but they don’t replace feel — momentum on the bench, how players are skating, or whether a team still believes it can win. In that sense, Roy’s stance isn’t old-school stubbornness. It’s selectivity. He knows the numbers. He just refuses to let them override what’s happening on the ice.
