> ## Documentation Index
> Fetch the complete documentation index at: https://offthepace.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Can a tyre recover once it is past the cliff?

> Post-cliff tyre behaviour by compound: how often a worn tyre shows pace recovery when the driver backs off, and whether the wear is thermal (recoverable) or structural (not).

Once a tyre is past the degradation cliff, is it gone, or can the driver nurse some pace back by lifting off? The recovery forecast answers that by looking only at post-cliff laps and asking how often, and how strongly, a tyre shows a partial recovery and critically, whether the wear is thermal (surface, recoverable) or structural (bulk, not).

## What it shows

The chart breaks post-cliff recovery into three signals per compound:

* **Recovery rate (`recovery_flag`)**: the fraction of post-cliff laps where the model flags any partial pace recovery. This runs very high across all compounds (\~86–89%) because the flag fires on any positive deviation from the cliff slope, however small.
* **Average recovery probability**: the model's probabilistic estimate of how likely a lap with the observed surface/bulk load split is to recover. This is far more discriminating than the binary flag typically 21–29%.
* **Surface/bulk ratio**: the ratio of surface thermal load to bulk mechanical load. Below \~0.4 the wear is **bulk-driven** (compound breakdown, hard to recover); above \~0.4 it is mixed or **surface-driven** (thermal, can partially recover with reduced push).

<Note>
  Only post-cliff laps are included the feature is silent about pre-cliff behaviour, which the [degradation timeline](/app/stint-degradation-timeline) covers.
</Note>

## How to use it

1. **Select a compound** to see its post-cliff recovery profile.
2. **Trust the probability over the rate** the near-universal recovery rate is a low bar; the average recovery probability is the number that separates compounds.
3. **Read the surface/bulk ratio** to know *why* a tyre can or cannot recover: a low ratio means structural breakdown that lifting won't fix.
4. **Cross-reference a real stint** in the [pit strategy Gantt](/app/pit-strategy) to see whether a driver who extended was riding recoverable thermal wear or terminal structural wear.

## Reading the signal correctly

A high recovery rate does **not** mean tyres bounce back to fresh pace it means a measurable, often tiny, deviation above the cliff slope. The magnitude lives in the average recovery probability, not the flag.

<Warning>
  Surface/bulk ratios near the \~0.4 boundary are ambiguous: the same lap can read as thermal or structural across nearby laps. Treat the boundary as a soft transition, not a hard switch.
</Warning>

## Data source

The chart queries `int_tyre_surface_vs_bulk_decoupling` via DuckDB-Wasm in the browser, using post-cliff laps only.

| Column                 | Meaning                                                                |
| ---------------------- | ---------------------------------------------------------------------- |
| `recovery_flag`        | Whether a lap shows any positive deviation from the cliff slope        |
| `recovery_probability` | Model probability the lap genuinely recovers, averaged per compound    |
| `surface_bulk_ratio`   | Surface thermal load ÷ bulk mechanical load (the recoverability split) |

The data window covers **2018–2024**, post-cliff laps only.

For the full model definition, see the [`int_tyre_surface_vs_bulk_decoupling` reference](/reference/models/int/int_tyre_surface_vs_bulk_decoupling).