← All posts
Model Explainer4 min readData: 2026 season to date

Reading live tyre degradation: what our four tyre numbers mean

Tyre health, degradation rate, predicted cliff lap, and cliff risk — the four tyre-state outputs RaceHooks publishes every lap, and how to read them during a live race.

Every lap, a Formula 1 tyre is a little slower than it was the lap before — and the hard part isn't knowing that, it's knowing how much, and knowing it before the lap times make it obvious. By the time a driver is visibly struggling, the strategist is already late. So RaceHooks publishes four tyre numbers on every lap, built to answer the question a pit wall actually asks: how much longer can this set last, and what does it cost to stay out?

The four numbers we publish

These arrive as part of the analytics layer on every lap, per driver:

OutputRangeWhat it tells you
Tyre health0.0–1.0How much usable performance is left in the set right now
Degradation rateseconds / lapHow much lap time the tyre is losing per additional lap
Predicted cliff laplap numberThe lap where degradation is expected to turn sharply non-linear
Cliff risklow / medium / high / criticalA banded, at-a-glance read of how close the cliff is
0.0–1.0
the tyre health scale we publish for every car, every lap

Tyre health is the headline figure, but it's the rate and the cliff that drive decisions. A car at 0.45 health losing 0.05s a lap is in a very different situation from a car at 0.45 health losing 0.30s a lap, even though the health number is identical.

Why raw lap times lie

The obvious way to measure degradation is to watch a driver's lap times climb. It doesn't work, because three other forces are moving lap time at the same time:

A raw lap-time delta blends all of these together. Our tyre-state model exists to separate the tyre's contribution from everything else, and update that estimate as each lap completes — so the degradation rate you read is the tyre, not the fuel load or the track.

Why we don't just publish lap times

Anyone can read a timing screen. The value isn't the lap time — it's the part of the lap time that belongs to the tyre, isolated in real time. That's the number a strategy call actually turns on.

What "the cliff" actually is

Tyre degradation isn't linear. For most of a stint a tyre loses performance gently and predictably. Then it reaches a point — the cliff — where the surface or the structure gives up and lap time falls away fast. Crossing the cliff can cost more in a handful of laps than the entire preceding stint did.

That's why the predicted cliff lap is a separate output from the degradation rate. A gentle, steady rate can still be sitting two laps short of a cliff, and a strategist who only watches the rate won't see it coming. We model the cliff explicitly and publish the lap we expect it on, so the question becomes "do we pit before lap N?" rather than "does this feel like it's about to go?"

Reading cliff risk in real time

Cliff risk is the field built for glance-speed decisions. It collapses the health, the rate, and the distance-to-cliff into one banded signal:

The point of banding it is speed. During a live session nobody has time to integrate three continuous numbers in their head every lap; they need a signal that says act now the moment it's true. That's what cliff risk is for — and it's one of twelve production models running on every RaceHooks session.

The takeaway

Tyre health tells you what's left; the degradation rate and the predicted cliff tell you what it costs to stay out — and cliff risk tells you the instant that math flips.

RaceHooks is a motorsport analytics platform — twelve production ML models enriching live race data, delivered to you in real time.