RaceIQ · Formula 1 2026
Read the grid before lights out
Open-source machine-learning forecasts for every Grand Prix — backtested across 48 races and graded against every official result.
Track record
Graded against every result
Every forecast is scored against the official classification once the chequered flag drops — no cherry-picking. The figures below are out-of-sample: the model never saw these races during training.
This season, live: 57.1% podium & points accuracy — how often the model puts the right drivers on the podium and in the points — across 7 rounds graded so far.
How it works
Live data → model → forecast
From raw timing sheets to a probability for every car — here is the path each forecast travels before it reaches you.
Step 1 / 3
Live F1 data
- 1
Live F1 data
Race results, telemetry, weather and standings ingested every Grand Prix weekend.
- 2
The model
A learning system that re-trains as the season progresses, then weighs every car in the field.
- 3
Race forecast
Probabilities for win, podium, top six and full classification — refreshed before lights out.
What you get
Every read, one tap away
Under the hood
Built to be checked
No black box, no cherry-picked highlights. The system is automated, time-stamped and open — so you can verify every claim yourself.
Re-trained every weekend
An automated pipeline folds in the latest race before the next Grand Prix — recent form, upgrades and weather, no manual intervention.
Time-stamped forecasts
Every forecast carries the moment it was generated, so freshness is never ambiguous. Last refreshed Jun 15, 2026.
Open methodology, open source
The full pipeline and the exact accuracy scoring are public on GitHub — every number on this site is reproducible.
Decades of racing behind it
Grounded in 943 graded driver-rows, with historical data reaching back to 1950.
Questions
Good to know
Is this betting or gambling advice?
No. RaceIQ is a personal project published for education and entertainment. The forecasts are model outputs and should not be used for betting or any form of gambling. The project is not affiliated with Formula 1, the FIA, or any constructor.
How accurate is it, really?
Every predicted finishing position is graded against the official classification once a race is over — a hit lands within three places of the actual result. The backtested and live numbers are published in full on the accuracy report, including the rounds where the model got it wrong.
How fresh are the forecasts?
Forecasts are regenerated every Grand Prix weekend and time-stamped, so you always know how recent a prediction is. “Live” means a session — practice, qualifying or the race — is happening as you read.
Is it open source?
Yes. The full data pipeline, the model, and the exact accuracy scoring are public on GitHub, so every figure on this site is reproducible.
What data feeds the model?
Race results, session telemetry, weather and championship standings, ingested each weekend — grounded in historical results reaching back to 1950.
Lights out
Ready before the grid forms
Open the next Grand Prix forecast — then judge it against the result. That is the whole deal.