Hockey results Friday 23 October 2026: AI predictions review

Six NHL games, ten frontier models and one clear headline: the TuringStats consensus went 4 of 6 on Friday 23 October 2026. Four winners were called correctly, two were not, and the scorelines the models favoured were close but rarely exact.

The biggest surprise came in St. Louis, where all ten models backed Vegas Golden Knights and the Blues answered with a 6–2 rout. Seattle delivered the second shock, beating Utah Mammoth 5–3 despite 90% of models favouring the visitors.

Full-time scenes: a six-game NHL slate produced four correct consensus calls and two upsets.

Where the AI got it right

Dallas was the day's most unified call. All ten models backed a Stars win and the most common scoreline was 3–2; Dallas Stars 3–2 Winnipeg Jets landed exactly on that number. Average confidence sat at 60%, modest for a unanimous pick, but the verdict and the score both held. Dallas Stars vs Winnipeg Jets prediction

Calgary drew only 70% of the ten-model vote, yet the Flames edged Minnesota Wild 2–1. The consensus scoreline of 3–2 was one goal off in total but the winner was right, and average confidence of 55% reflected a genuinely split panel. Calgary Flames vs Minnesota Wild prediction

Edmonton produced the night's highest-scoring game and the models stayed with it. Seventy percent backed the Oilers, the most common score was 4–3, and Edmonton Oilers 6–5 Carolina Hurricanes delivered the winner in a ten-goal thriller. The scoreline missed badly; the result did not. Edmonton Oilers vs Carolina Hurricanes prediction

Vancouver was the fourth hit. Eighty percent of models favoured Detroit Red Wings and the visitors won 5–2, comfortably clear of the 3–4 scoreline the panel had settled on. Average confidence of 58% made this one of the firmer calls of the day. Vancouver Canucks vs Detroit Red Wings prediction

Where it got it wrong

St. Louis Blues 6–2 Vegas Golden Knights was the day's heaviest miss. Every one of the ten models picked Vegas, the most common score was 3–4, and average confidence reached 56%. The Blues scored six and conceded two, so neither the winner nor the margin survived contact with the result.

Seattle Kraken 5–3 Utah Mammoth was the more costly in confidence terms. Nine of ten models backed Utah, the consensus score was 4–5, and average confidence of 61% was the highest of any match on the card. The Kraken won by two, and the panel's strongest conviction produced nothing.

Models of the day

Five models tied for the best tally at 4 of 6: GPT-5.6 Luna, Claude Sonnet 5, MiMo V2.6 Pro, Grok 4.6 and GLM 5.3. All five avoided both upsets and matched the overall consensus record exactly.

Three models finished a game behind at 3 of 6: DeepSeek V4 Pro, Qwen 3.8 Max and Kimi K3. At the bottom, Mistral Medium 3.5 and Gemini 3.8 Flash both returned 2 of 6, each losing four of the six calls. No model on the card beat the consensus, and the spread between best and worst was two correct results.

Full results

• NHL: Dallas Stars 3–2 Winnipeg Jets — consensus Dallas Stars win (100%) RIGHT
• NHL: St. Louis Blues 6–2 Vegas Golden Knights — consensus Vegas Golden Knights win (100%) WRONG
• NHL: Calgary Flames 2–1 Minnesota Wild — consensus Calgary Flames win (70%) RIGHT
• NHL: Edmonton Oilers 6–5 Carolina Hurricanes — consensus Edmonton Oilers win (70%) RIGHT
• NHL: Seattle Kraken 5–3 Utah Mammoth — consensus Utah Mammoth win (90%) WRONG
• NHL: Vancouver Canucks 2–5 Detroit Red Wings — consensus Detroit Red Wings win (80%) RIGHT

Friday was a reminder that unanimity is not accuracy: the two calls every model or nearly every model agreed on split one right and one wrong. The next slate is already through the models. Full picks and confidence numbers are on the today's hockey predictions page.

Next up: today's hockey predictions from the same panel, and the results archive with every graded call. Forecasts are informational only.