Tyre degradation

Stint by stint

import fastf1
from lapbox.tyres import compound_summary, stint_degradation

session = fastf1.get_session(2025, "Monza", "R")
session.load(telemetry=False, weather=False, messages=False)

stints = stint_degradation(session.laps)
print(stints[stints["driver"].isin(["VER", "NOR"])].round(3).to_string(index=False))

driver            team  stint compound  laps  median_corrected  deg_per_lap
   VER Red Bull Racing      1   MEDIUM    36            81.327        0.006
   VER Red Bull Racing      2     HARD    15            80.989        0.032
   NOR         McLaren      1   MEDIUM    45            81.631        0.007
   NOR         McLaren      2     SOFT     6            80.970        0.043

Each stint’s timed, non-pit laps are fuel-corrected and kept within 107 % of the stint’s best; deg_per_lap is the slope of what’s left against lap number. A stint needs three such laps to appear, and a three-lap stint is a weak trend.

By compound

compound_summary takes the median slope per compound, so one driver’s ruined stint doesn’t define a compound:

print(compound_summary(stints).round(3).to_string(index=False))

compound  stints  laps  median_deg  best_deg  worst_deg
    SOFT       3    20      -0.039    -0.149      0.043
  MEDIUM      15   404      -0.001    -0.091      0.024
    HARD      16   489       0.005    -0.026      0.061

The per-compound model

TyreDegradationModel fits one straight line per compound over every driver: fuel-corrected lap time against tyre age. Clean the laps first — its 107 % filter catches safety-car laps, not every slow one:

from lapbox.pace import LapCleaningPipeline
from lapbox.tyres import TyreDegradationModel

model = TyreDegradationModel().fit(LapCleaningPipeline().run(session.laps))
print(model.summary().round(3).to_string(index=False))

compound  degradation_s_per_lap  intercept_s  fuel_coef    r2  laps
    SOFT                 -0.133       82.673      0.048 0.173    24
  MEDIUM                 -0.008       82.289      0.048 0.021   399
    HARD                  0.014       82.137      0.048 0.067   491

fuel_coef is the correction applied per lap of fuel (0.048 s, the same constant as everywhere else). intercept_s is the lap time on a new tyre with the fuel all burned.

print(f"{model.degradation_rate('HARD'):+.3f} s/lap")
print(f"{model.predict_time_loss('HARD', 20):+.2f} s after 20 laps")
hard = model.compounds["HARD"]
print(hard.laps, round(hard.r2, 3), round(hard.fuel_coef, 3))

+0.014 s/lap
+0.27 s after 20 laps
491 0.067 0.048

TyreDegradationModel.predict_time_loss is rate × age, against a fresh tyre. A compound that wasn’t fitted — fewer than five laps, or one tyre age — raises a KeyError.

How far to trust a rate

Both methods say the soft got quicker with age at Monza, on 20–24 laps. Neither is a measurement of the tyre alone: the slopes also carry track evolution, the fuel assumption and who ran which stint. The low r2 is the honest part: tyre age explains little of how the lap times varied here.

• Read every rate with its laps (and, for the model, its r2).
• Low-wear circuits give small, noisy slopes. Over 65 races of 2024–2026, 23 % of whole-race compound rates from the model still came out negative. lapbox reports the fit as it comes out; clamp it (at zero, say) before planning a strategy with it.
• Rain tyres mostly drop out of a mixed race. The 107 % filter keeps laps against the driver’s best of the race, usually on slicks: of eight rain-tyre fits in those 65 races, five went and two kept under a third of their laps. Fit rain tyres on their own laps.