← back to overview

Elo Ratings

Separating driver from car. An Elo rating system applied to the 2026 season to estimate driver skill independent of machinery.
Pairwise Elo K-factor 4 5 Races Processed DNF-Neutral

Key Metrics

Headline figures from the Elo model after Round 5

Final Elo Ratings

Horizontal bar chart after Round 5 — sorted highest to lowest, colored by team

Elo Progression

Round-by-round Elo movement from baseline 1500 through 5 races

Teammate Comparison

Elo gap within each team — the purest signal of driver vs. driver

TeamDriver 1EloDriver 2EloGap

The Math

How the Elo model works and what the numbers mean

Elo Rating System

The Elo system was originally developed for chess but applies naturally to any competition where you need to rank participants based on head-to-head outcomes. In F1, each race generates pairwise comparisons: every pair of drivers who both finish the race produces a "match" where the driver who finished ahead wins.

The Formula

Expected score: E_A = 1 / (1 + 10^((R_B - R_A) / 400))
Update rule:    R_A_new = R_A + K * (S_A - E_A)

where S_A = 1 if A finished ahead of B, 0 otherwise

Why K = 4?

In chess, K is typically 16-32 because each game is a single match. In F1, a 16-finisher race generates 16 * 15 / 2 = 120 pairwise comparisons per driver. A large K would cause wild rating swings after every race. K=4 produces stable, meaningful movement: a dominant race might move a driver +15 to +25 points, while a poor race costs a similar amount. Over a full season the ratings converge to reflect genuine performance differences.

What the Numbers Mean

DNF Handling

Drivers who retire (DNF) or do not start (DNS) are excluded from that race's pairwise comparisons entirely. They are not penalized — mechanical failures and crashes do not affect their rating. This is deliberate: Elo should measure driver skill, and a gearbox failure is not a skill signal.

Car vs. Driver

The fundamental caveat — and why teammate gaps matter most

Elo ratings in F1 still capture car performance to a significant extent. A driver in the fastest car will accumulate a higher rating not because they are necessarily more skilled, but because they finish ahead of more drivers in every race. This is an inherent limitation of any results-based rating system.

The teammate comparison is the purest signal available. Since teammates share the same car, the Elo gap between them isolates driver performance from machinery. A large intra-team gap suggests a genuine skill or consistency difference. A small gap suggests closely matched drivers.

Cross-team comparisons should be interpreted cautiously. If both Mercedes drivers sit at 1550+ while both Cadillac drivers sit at 1440, that tells you more about the cars than the drivers. But if one Mercedes driver is at 1580 and the other at 1530, the 50-point gap is a meaningful driver-skill signal — they had the same equipment.