Skip to main content
The era ratings timeline draws every driver’s career as a line of season ratings: how far ahead of (or behind) his teammate he was, in seconds per lap, in the same car. The car cancels in the teammate comparison, and so does the 2022 ground-effect regulation change, which moved both teammates alike, so a 2018 rating sits on the same scale as a 2025 one without any era correction. Negative means faster than his teammate.

​
What it shows

Each line is one driver’s season-by-season rating:
  • Rating (gap to teammate): per race, the median of the driver’s lap-by-lap gap to his teammate on laps both ran clean; per season, the mean of those. Negative = faster than his teammate(s).
  • Bayesian shrinkage: per-season means are shrunk toward the season mean with a 5-race prior. Drivers with fewer races have wider intervals and estimates pulled closer to zero; the uncertainty is honest, not hidden.
  • Line style: drivers with at least 8 races on each side of 2022 (“bridge” drivers) render as solid lines, others as dashed. This is context only; no era offset is applied and no rating depends on it.
  • 95% CI ribbon: the per-season posterior uncertainty.
A line moves when the driver’s margin over his teammate moves, and a change of teammate can move it as much as a change in the driver.

​
How to use it

  1. Select drivers to add or remove their career lines.
  2. Check the teammate at any step in a line: a jump often coincides with a new teammate.
  3. Compare ribbon overlap between two drivers in the same season: overlapping ribbons mean the gap is within noise.
  4. Follow a single arc across seasons to see development, peak, and decline on one continuous scale.

​
Reading the signal correctly

The rating is teammate-relative: it places a driver against the driver in the same car, not against the field and not on an absolute lap-time scale.
Until 2026-09-28 this page applied a “bridge driver” era offset to every pre-2022 season. It was not statistically distinguishable from zero and it inverted the gap it was meant to remove, so it was dropped (audit finding F45). The per-race statistic also changed then, from a best-laps ceiling against the teammate’s typical lap to a like-for-like lap-by-lap median (F40). See the decomposition limitations.

​
Data source

The chart queries int_era_normalized_driver_rating (built on int_driver_season_ratings) via DuckDB-Wasm in the browser; the clean-lap filter is applied upstream. The data window covers 2018–2025. For the full model definitions, see the int_era_normalized_driver_rating reference and the int_driver_season_ratings reference.