Connor McDavid Is On Pace to Break All The Records. Is it Sustainable?

Connor McDavid has 34 points in his first 12 games, a pace for 238 over this year's 84-game schedule, and nobody has ever scored more than 215. So I went through every season since 1917, found everyone who ever started like this, built a model that learns how much a hot start is real, and played out the rest of McDavid's season 10,000 times. Every player's analytics, from expected goals to 5-on-5 impact: https://spreadsheetsports.com/hockey/... Chapters 0:00 On pace for 238 1:00 The start (and why it isn't the best ever) 1:42 Adjusting for era 2:50 Every great hot start slows down 3:55 The model and the backtest 5:12 10,000 seasons 7:20 So, is it the greatest season ever? 8:28 Go deeper How the model works: every season since 1945 was frozen at the 12-game mark (the date the median team had played 12 games). Points per game are era-adjusted to 2025-26 scoring (league skater points per team-game), so 1980s totals and today's are on the same scale. A regression fitted on 1,238 star seasons (prior season 1.0+ adjusted points per game) learns how much the first 12 games predict the rest of the year: the start carries about a fifth of the weight and the previous three seasons about four fifths. The rest of McDavid's season is then simulated 10,000 times with his expected rate (1.76 points a game from here on), a spread for true talent, game-to-game randomness, and how often stars actually play their remaining games. The rest of the league gets the same model, so the race for second and the dominance numbers are simulated too. The backtest: for every season since 1960-61 the model was re-fit without that season, shown only the first 12 games, and asked to predict how each star finished. On the 87 hottest starts, simply keeping the pace missed the final total by 36 points on average (almost always too high); the model missed by 12 and was closer 83% of the time. Its biggest miss: Gretzky 1983-84 (predicted about 170, he scored 205). The results: McDavid's middle outcome is about 150 points (80% range 119 to 174). He beats his own career high of 153 in 44% of the simulations, reaches 160 in about 31% and 200 almost never. If he plays every remaining game, the middle outcome is 161 and he beats 153 about 68% of the time. Era-adjusted, there is about a 1 in 14 chance (7%) he beats Gretzky's 1983-84 rate and a 1 in 26 chance (4%) he beats Lemieux's best, which would make it the best season since 1945. Matching Gretzky's 1.57x dominance over the next-best scorer happens about 3 times in 1,000, mostly because today's runner-ups (up to 1.7 adjusted points a game) are far better than the 1980s ones (about 1.4). Caveats: numbers are through 12 games (Oct 26 2026). The model doesn't know about the new McDavid-Draisaitl line, and it treats the season's remaining scoring environment as normal. Edmonton is shooting 16.6% (league average 11.4%) while being outshot, which the model handles only through the small weight on the start. Sources: NHL stats API (skater and team summaries, game logs, every season 1917-18 to 2025-26); TSN (fastest to 20 assists, Oct 21 2026); The Hockey News. Footage and audio: TSN SportsCentre (Craig Button, Oct 23 2026), TSN OverDrive (Feb 1 2023), Daily Faceoff (Cameron Gaunce, Oct 24 2026), NHL (Gretzky's 51-game point streak; Lemieux, Dec 31 1988; Oilers vs Flames, Dec 27 2025), NHL Network (Aug 4 2023), SPORTSNET (McDavid's 150th point, Apr 8 2023; McDavid 2025-26 and 2022-23 highlights; Mike Bossy tribute), The Canadian Press (Wayne Gretzky, Oct 18 2016). Does he get to 160? Tell me in the comments. #NHL #hockey #ConnorMcDavid #Oilers #Gretzky #Lemieux