Our odds

Before each grand prix we rate every driver for the race ahead and turn those ratings into win, podium, top six and top ten probabilities. The model is built mainly from what the cars have done on track in the recent past, tempered by a season-form prior from the championship standings so a proven front-runner's single bad weekend doesn't sink them. It does not try to second-guess strategy or weather.

What the model reads

The sample is the six most recent sessions that carry lap data — about three race weekends — taken from races and qualifying only. Each one is weighted by two things: how recent it is, and how closely its circuit resembles the track coming up. The newest session counts for most.

Circuit similarity then shifts the balance. A technical, high-downforce lap tells you more about another technical track than a street circuit does, so a similar circuit in the sample is trusted more than a dissimilar one. The two weightings combine and are normalised, so the sessions always add up to a full picture rather than double counting.

The six factors

From the cleaned laps and the classified results, each driver earns a rating built from six groups of signals:

• Pace. Median race-lap time against the field's best, lap-time consistency, sector times and speed-trap figures. This is the largest part of the rating. Pace is also shrunk toward each team's car baseline, so a thin or disrupted sample leans on the machinery rather than noise.
• Reliability. Finish rate across the sample, smoothed toward the field's finish rate so a single retirement dents a score rather than wrecking it.
• Pit performance. Pit-lane times and how consistent they are, with a small penalty for cars that have to stop more often than the field.
• Race craft. Places gained between the starting grid and the chequered flag.
• Recent form. Finishing positions, leaning towards the latest race, with a season-long prior from the championship standings so proven pace tempers a rough patch.
• Grid. A boost for starting near the front, where track position is hardest to overturn.

From ratings to probabilities

A team-relative adjustment comes first, nudging the quicker driver in a pairing ahead of their team-mate. That step feeds the probabilities only; the rating we display stays as the raw pace-based figure.

Win probability comes from a softmax, which converts the ratings into shares that sum to one while keeping a clear gap between the favourite and the rest. Podium, top six and top ten come from a Monte Carlo simulation: the model runs eight thousand races, drawing a finishing order each time from the same weights, and a driver's probability is the share of those races in which they land in the bracket. The simulation is seeded, so the same data always returns the same numbers.

What the numbers are, and are not

These are estimates of recent pace, not forecasts of the result. The sample is short and weighted hard towards the last race, so the model can read the form differently from the championship table. A driver who retired from the most recent and highest-weighted race can sit low here even while leading the title, simply because the model saw no race pace from them that weekend.

Treat the figures as a snapshot of who has been quick lately rather than a verdict on Sunday. You can change the assumptions yourself in the interactive calculator, adjusting how much each factor counts and watching the probabilities move. Each grand prix also has its own odds page, with the current grid's past results at that event.

Disclaimer

The odds on this site are a data-derived estimate generated from public timing data. They are not betting advice and they are not a prediction of any outcome. Pit Lane Data is independent and not affiliated with any bookmaker. If you choose to bet, that is your decision and your responsibility. Betting is for over-18s only; please gamble responsibly. Free, confidential support is available at BeGambleAware.org.