Models / BiLSTM

Pit Stop Prediction

P(this driver pits this lap), every lap.

BiLSTManalytics.strategy

How it works

A bidirectional LSTM reads a 20-lap rolling feature window per driver — tyre age, fuel-corrected pace delta, stint length, gaps to the cars ahead and behind, and track status — and emits a pit probability each lap. A recommendation flag trips when the probability clears an era-adjusted threshold (0.53 pre-2026, 0.60 for 2026).

Inputs

  • 20-lap feature window
  • tyre age & compound
  • fuel-corrected pace delta
  • gaps ahead/behind
  • track status

Output

drivers[].pitStopProbability0–1 probability of a stop this lap.
drivers[].pitRecommendedtrue when the probability clears the era threshold.

Calibration

Era-specific thresholds; recalibrated as each season's stop patterns accumulate. See the calibration reference for the shared pipeline.

← All models