How We Grade F1 Drivers: Season Performance vs. Teammate Dominance

Every driver on our Formula 1 dashboard receives a dual-metric rating, updated automatically after every Grand Prix. To provide a complete and mathematically rigorous picture of a driver's performance, we evaluate them on two fronts: the Season Performance Score (SPS) for absolute results, and the Teammate Score (TMS) for isolated driver talent.

Why Do We Use Two Scores?

In Formula 1, roughly 70–80% of a race result comes down to the car, not the driver. A highly talented pilot in a Williams will rarely outscore a mediocre one in a Red Bull. Because of this, any single number that tries to combine car dominance and driver talent is ultimately misleading.

Our solution is to separate them:

The Season Performance Score (SPS) reflects what a spectator observes on TV—points, podiums, and consistency. It intentionally includes the "car effect" because that is the reality of the World Championship standings.
The Teammate Score (TMS) isolates the driver's individual talent by comparing them to the only person on the grid driving the exact same machinery: their teammate.

Read them together, and you get a complete and fair assessment of a driver's true performance.

The Science Behind the Weights

The weightings in our formulas are not arbitrary. They are the result of statistical calibration using Ridge regression on a dataset of 48 reference driver-seasons (ranging from Ayrton Senna in 1988 to Max Verstappen in 2024). Each historical season was assigned a target score based on historical consensus, and the regression determined which metrics best predict actual performance.

Our analysis revealed that Championship points (P1) are the strongest single predictor of performance. Qualifying (P2) provides a vital complementary signal for pure speed. A fourth tested pillar—detailed race results like wins and podiums—was ultimately removed. It showed a 0.97 correlation with championship points, meaning it added almost no new statistical information. Removing it simplified the model without sacrificing accuracy.

1. The Season Performance Score (SPS)

$SPS = 0.50 \times P1 + 0.25 \times P2 + 0.25 \times P3$

The SPS measures overall performance across the season. The score is the weighted average of three independent pillars, each normalized from 0 to 100.

P1 — Championship (50%)

$P1 = 0.70 \times pts\_relative + 0.15 \times pts\_absolute + 0.15 \times teammate\_ratio$

Measures how the driver performs in the championship fight using three complementary components:

relative position to the season leader

$pts\_relative = \frac{driver\_points}{best\_season\_score} \times 100$

absolute yield against the theoretical maximum

$pts\_absolute = \frac{driver\_points}{25 \times num\_races} \times 100$

dominance over the other side of the garage

$teammate\_ratio = \frac{\text{clamp}(driver\_points / teammate\_points, 0, 3)}{3} \times 100$

P2 — Qualifying (25%)

$P2 = 0.55 \times grid\_position\_score + 0.45 \times h2h\_combined$

Pure speed on Saturday. Combines average grid position (absolute pace) with the percentage of qualifying sessions where the driver beat their teammate (relative pace).

$grid\_position\_score = \frac{20 - avg\_grid\_position}{19} \times 100$

$h2h\_combined = \frac{(h2h\_quali \times n\_quali) + (h2h\_shootout \times n\_shootout)}{n\_quali + n\_shootout}$

P3 — Reliability & Consistency (25%)

$P3 = 0.50 \times (finish\_rate \times 100) + 0.50 \times (points\_scoring\_rate \times 100)$

Measures the driver’s ability to see the checkered flag and score points. This is highly independent; a driver can be fast (high P2) but accident-prone (low P3).

$finish\_rate = \frac{num\_races - num\_dnf}{num\_races}$

$points\_scoring\_rate = \frac{races\_in\_points}{num\_races}$

SPS Tiers:

S ≥ 90

A 78 – 89

B 65 – 77

C 50 – 64

D 35 – 49

F < 35

2. The Teammate Score (TMS)

The TMS measures head-to-head dominance. Because both drivers share the exact same car, differences here isolate raw talent.

The S-Curve Model

All head-to-head (H2H) ratios pass through a logistic S-curve function. This mathematical model amplifies clear dominations and smooths out matches that are close to 50/50.

Actual Head-to-Head	Score After S-Curve	Interpretation
95% / 5%	~97	Overwhelming domination
75% / 25%	~80	Clear superiority
60% / 40%	~60	Slight advantage
50% / 50%	50	Perfect equilibrium
25% / 75%	~20	Clearly dominated

The TMS Formula

With Sprint Races:

$TMS = 0.45 \times D1 + 0.35 \times D2 + 0.20 \times D3$

Without Sprint Races:

$TMS = 0.55 \times D1 + 0.45 \times D2$

D1 — Race Pace

$D1 = 0.40 \times ratio\_points + 0.35 \times h2h\_finish + 0.25 \times avg\_gap$

This pillar captures race dominance against the teammate using three sub-components:

the points ratio (which reflects the value of the results)

$ratio\_points = \text{S\_curve} \left( \frac{driver\_pts}{driver\_pts + teammate\_pts} \right)$

the head-to-head finishing record (which counts every race equally)

$h2h\_finish = \text{S\_curve}(\% \text{ races finished ahead})$

the average gap in finishing positions (which measures the sheer scale of the dominance)

$avg\_gap = \text{clamp}(50 + delta\_positions \times 6, 0, 100)$

D2 — Qualifying Pace

$D2 = 0.55 \times h2h\_quali + 0.45 \times grid\_gap$

This measures raw speed and domination over a single lap. It combines:

standard qualifying sessions

$h2h\_quali = \text{S\_curve} \left( \frac{h2h\_combined}{100} \right)$

Sprint shootouts into a single head-to-head metric (weighted by the number of sessions), alongside the average gap in starting grid positions

$grid\_gap = \text{clamp}(50 + delta\_grid \times 10, 0, 100)$

D3 — Sprint Dominance (Active during Sprint weekends only)

$D3 = 0.60 \times h2h\_sprint + 0.40 \times ratio\_sprint\_pts$

Active only during Sprint weekends, this compares:

sprint finishing positions

$h2h\_sprint = \text{S\_curve}(\% \text{ sprints finished ahead})$

points scored between the two teammates

$ratio\_sprint\_pts = \text{S\_curve} \left( \frac{driver\_sprint\_pts}{total\_team\_sprint\_pts} \right)$

If no sprints have taken place in the season, this pillar is deactivated and its 20% weight is dynamically redistributed to Race (D1) and Qualifying (D2).

TMS Tiers:

S ≥ 80

A 65 – 79

B 55 – 64

C 45 – 54

D 35 – 44

F < 35

3. Reading the Data: Examples in Action

To understand how the math translates to the track, here is how the dual-score system evaluates different scenarios:

Max Verstappen (2024)

SPS: 94.4 (S) · TMS: 92.8 (S)

A dominant season in a strong car. He beat Perez 96% of the time in qualifying and 100% in races. Both scores converge to tell a simple story: an elite driver in an elite car.

Fernando Alonso (2025)

SPS: 42.5 (D) · TMS: 79.9 (A)

Stuck in a weak car (51 points total), but utterly dominated Stroll 100% in qualifying and 82% in races. The low SPS reflects the poor machinery, while the high TMS isolates Alonso's individual talent in maximizing it.

Lando Norris / Oscar Piastri (2025)

Norris SPS: 82.3 (A) · TMS: 47.0 (C)
Piastri SPS: 81.8 (A) · TMS: 53.0 (C)

A strong McLaren car piloted by a highly balanced duo (55/45 race split, 54/46 qualifying split). Their SPS ratings are near-identical, and their TMS numbers neutralize perfectly around the 50-mark. Neither clearly dominates the other.

Lewis Hamilton (2024)

SPS: 56.0 (C) · TMS: 27.5 (F)

Driving a decent Mercedes (207 points, 2 wins), but heavily outperformed by Russell—beaten 79% of the time in qualifying and 65% in races, finishing an average of 1.4 positions behind. This yields a mid-table SPS, but a poor TMS reveals the teammate is doing much more with the same car.

4. Input Data & Methodology

The dashboard relies entirely on official, objective results available immediately after a Grand Prix. No subjective ratings are used.

Data Used: Race (positions, points, DNFs), Qualifying (positions), Sprint Qualifying/Shootouts (positions), Sprint (positions, points).
Data Ignored: Free practice times, post-qualifying grid penalties (we use raw qualifying pace), and individual lap times.

5. Context & Limitations

The Car Effect is Intentional (SPS): A highly talented driver in a backmarker car will score poorly in the SPS. This reflects the reality of the standings. The TMS exists to provide balance.
Mechanical vs. Driver DNFs: The formula currently treats all Did Not Finishes (DNFs) equally. Distinguishing between a mechanical failure and a driver error requires subjective annotation, which falls outside our automated methodology. All DNFs penalize the P3 (Reliability) pillar.
Teammate Strength is Not Adjusted: Beating Max Verstappen is harder than beating a rookie. Without a complete transitive grid network, the TMS measures domination over a specific teammate, not absolute grid-wide talent.
Early Season Volatility: Ratios are normalized by the number of races completed. After just 3 races, a single DNF causes massive score swings. Scores reflect what has happened, not a year-end projection, and stabilize naturally around mid-season.