My Model Predicts the 2027 Stanley Cup Champion

Spreadsheet Sports builds a preseason Stanley Cup model and keeps only what survives a backtest. Every idea is tested on 18 seasons (2009 to 2026), each time using only what was known before that season started, and judged by one thing: did the model’s favourite win the Cup? The bar is the betting markets, whose preseason favourite won 3 times in those 18 seasons.

The versions tell the story. Last season’s standings picked 2 champions. Expected goals, which measure the chances a team created and gave up rather than the goals that went in, jumped to 5. Then two things every fan swears by made it worse: rebuilding rosters for summer trades and signings, and crediting teams for their goaltending, which turns out to be the most random thing in hockey. A “lab” threw out the Presidents’ Trophy curse, the Cup hangover, special teams and the hot goalie. Two ideas survived: two seasons of expected goals instead of one, and playoff experience, counted as series won over the last three years. With those, the favourite won 6 of 18 Cups, and on average the model gave the real champion a higher chance than the markets did.

Simulating 2026-27 ten thousand times, Carolina comes out on top at 22.8%, Colorado leads the West, and a Canadian team ends the drought in about one season in seven. He is careful to say the favourite still loses most of the time, and he promises to come back in June and grade every pick.