Reference

Models Reference: Motorsport Prediction Models

How the intelligence is built. Fourteen production ML models, plus an algorithmic layer (Monte Carlo win/podium probability with a full position distribution, ECP/ECPA expected championship points, and EKF tyre health), produce the analytics.* and weather.* feeds. This page documents how each model works: its type, inputs, and calibration. For the payload each one emits, see the Analytics feeds →

ℹ

Models are delivered as feed outputs, not as a separate product. You subscribe to a feed (e.g. analytics.strategy); the models behind it run server-side in real time. The model count is the documented source of truth; the live GCS manifest is the runtime truth.

Production ML models

Model	Type	Predicts	Feed
pit stop probability	gradient-boosted-ensemble	Probability of a green-flag pit stop on the next lap; stops under safety car, VSC or red flag are not counted.	analytics.strategy
safety car	gradient-boosted-ensemble	P(SC or VSC deployment next lap), session-level.	analytics.strategy
overtake	gradient-boosted-ensemble	P(a clean on-track pass on the next lap), per driver.	analytics.strategy
LTOE (Lap Time Over Expected)	gradient-boosted-ensemble	Lap Time Over Expected — the normalized pace delta, per lap.	analytics.strategy
pace mode	calibrated-classifier	Which pace mode a driver is running — push, hold, manage, warm-up or delta — per lap.
tyre cliff	gradient-boosted-ensemble	P(tyre performance cliff), per driver per lap.	analytics.strategy
undercut	calibrated-classifier	P(pitting now puts the driver ahead of the car in front once it covers), when within 5s — plus WHEN to strike: the recommended undercut lap and its success probability.	analytics.strategy
overcut threat	gradient-boosted-ensemble	P(staying out longer beats the car ahead's stop), when in range.	analytics.strategy
qualifying pace	gradient-boosted-ensemble	Per-sector pace deltas vs the session median, per driver.	analytics.qualifying
fastest lap	gradient-boosted-ensemble	P(this driver sets the session's fastest lap).	analytics.race-odds
DNF risk	survival-model	Retirement hazard per driver, per lap.	analytics.strategy
tyre compound	gradient-boosted-ensemble	The next compound a pitting driver takes, as probabilities.	analytics.tire-strategy
pit-stop count	gradient-boosted-ensemble	P(one / two / three+ stops) for the race, per driver — re-estimated every lap from the race so far.	analytics.strategy
pit-window survival	survival-model	How long each driver's current stint will last — expected laps to the stop and P(pit within 3 / 5 laps), given the stint's age.	analytics.strategy

Every model has its own page; click a name. All share the training and calibration pipeline below.

Algorithmic intelligence layer

Three non-ML components ship alongside the models and produce the bulk of the race-odds and tyre signals.

Win probability: Monte Carlo

A forward Monte Carlo race-to-finish simulation produces a full 20-element positionDistribution per driver, from which podium/points odds, expected points, expected laps led, and forward position-change projections are derived analytically. Powers analytics.race-odds and analytics.team-points (via joint-distribution convolution). Overview: the F1 probability API.

Tyre health: EKF

An extended Kalman filter tracks each driver's lap-delta series to estimate pace.tyre.health (0–1), degradation rate (s/lap), and stint lap. It also holds up against safety-car and traffic noise that fools a naive regression.

Expected championship points: ECP/ECPA

Season-level points expectation (and an adjusted variant) projecting championship outcomes from the current standings and per-race position distributions. Surfaces on analytics.championship-probability.

Calibration

Calibration is assessed on held-out sessions. A published 0.30 is a model estimate of a 30% chance, not a guarantee that the event will occur 30% of the time in every race state. Check each model reference for its horizon, eligible states and limitations. Current artifact and qualitative calibration status are available through the model manifest endpoint. The curated demo race exposes Analytics outputs on any plan. Ordinary replays follow plan feed entitlements; Free replay access does not include models for every session. For wider evaluation data, discuss a Custom engagement.

Changing regulations and model status

Changes in regulations, tyres and race operations can shift the relationship between inputs and outcomes. Evaluate the model version and period relevant to your workflow. Historical adjustment notes describe methodology; they are not a current production-health report. The model manifest is the reference for current artifact and calibration status.

← Analytics feeds API Reference →