CANADIAN GP · RECORDED MODEL RUN · 15 OCTOBER 2026
From race context to a reviewable decision.
Two concrete examples from a production-recorded replay of the 2026 Canadian Grand Prix. Explore what the models returned, how an analyst could use it and what this evidence establishes.
Recorded model run, 15 October 2026. These outputs were generated by replaying the race after the event. Use them to explore the product; assessing predictive performance requires a separate evaluation.
TRADING CASE · LAP 18
An independent probability view alongside your prices.
The race-to-finish model returns a finishing distribution. The snapshot below shows the four highest winner probabilities in the recorded frame, with podium probability and expected championship points from that same frame.
| Driver | Win | Podium | Expected points |
|---|---|---|---|
| George RUSSELL | 36.8% | 76.3% | 17.69 |
| Kimi ANTONELLI | 28.6% | 71.9% | 16.67 |
| Max VERSTAPPEN | 12.9% | 51.9% | 13.79 |
| Lewis HAMILTON | 10.6% | 46.3% | 13.57 |
Review workflow: align this timestamp and race state with your own prices, then inspect changes around pit stops and caution periods. Decide in advance how you will score eligible forecasts and handle missing states.
Generated 2026-10-15T18:46:03.480Z · feed analytics.race-odds · horizon: race finish. Win is positionDistribution[0]. This selected snapshot contains no market prices, settlement result or profitability comparison.
Request a trading-model walkthrough →TEAM CASE · LAP 18
Compare rival stop risk with strategy alternatives.
At the lap-18 checkpoint, Verstappen’s next-lap green-flag pit probability was 10.2%. The derived three-lap probability was 27.7%. Different horizons answer different questions; neither is a commitment that a stop will happen.
In the strategy output for Antonelli, the emergent baseline returned 16.88 expected points. The first listed alternative returned 17.89 expected points, with proposed stops on laps 23 and 43.
Review workflow: use the rival forecast as an input to scenario review, then compare plans with your team’s constraints and assumptions. The difference between simulated expected points is not a measured gain on track, and a recorded alternative is not an instruction to pit.
Feeds analytics.strategy and strategy.recommendation · session 2026-canada_r · last recorded lap-18 frames. Next-lap pit forecasts exclude stops under SC, VSC and red flag. Pit forecast generated 2026-10-15T18:46:03.480Z; strategy comparison generated 2026-10-15T18:44:51.182Z. These are separate recorded frames.
Discuss your strategy workflow →Bring your own evaluation criteria.
The recorded outputs let you review the probability definitions, strategy alternatives and the way they fit your workflow. For a performance evaluation, agree the dataset, forecast timestamps, model version, comparison baseline and scoring method with us.
These are examples from one race. Expected points describe model-estimated outcomes; they are not points gained on track. RaceHooks outputs support analysis and are not official results or settlement authority.
Read model definitions and limitations →