Methodology

P2P Driver Grades

We evaluate drivers across six dimensions: Pace (35%), Racecraft (20%), Qualifying (20%), Tire Management (10%), Discipline (10%), and Consistency (5%). Where a grade compares teammates, it isolates driver performance from the car by looking at two drivers in the same machinery. Pace is a blend: 55% where the driver ranks across the whole field, which mostly reflects the car, and 45% the gap to their teammate, the only part that separates driver from car. Scores update after every race weekend. Per-race breakdowns show how a driver's performance evolves across the season, and momentum indicators highlight current form trends. Grades draw on lap time telemetry, qualifying deltas, tire degradation rates, and FIA steward decisions.

A driver needs at least 3 graded races before we publish a grade at all, and Consistency specifically needs at least 5, since a race-to-race spread computed from fewer than that isn't a reliable read on volatility versus a genuinely unlucky small sample. Below those bars we show a neutral value rather than a confident-looking number we can't back up.

Racecraft is built lap by lap from three things: passes made on track (worth more inside the points and at circuits where passing is hard), defending a position while a car is close behind, and the opening lap. Gains made in the pits, passes into a car that was stopping, and places won back within five laps of a bad start do not count. Each race is scored against that race's own field, with extreme single-race scores clipped so one unusually pressured drive cannot set the scale, and the season grade is the average of those race grades. Losing several places in one lap counts as a mistake, except when the car was failing: losses in the last two laps before a retirement, or in a round whose retirement we researched as mechanical, are ignored.

A driver graded in fewer than 75% of the rounds they entered, because early retirements and non-starts left too little of the season to grade, is shown below the ranked field as NR with their grades and the reason. They still count in tier tables and best-of lists; only the rank is withheld. The grade describes how the driver performed when the car was running, so it says little about reliability.

Known limits, stated plainly: Discipline caps a completely clean driving record at 92, not 100, because stewarding data alone can't prove a flawless season, only the absence of a recorded incident. Tire Management shows “n/a” at circuits where tyre wear is naturally too low to produce a measurable degradation signal (Monza and Baku, most seasons) rather than guessing. And Consistency can't tell a genuinely erratic driver from one who had a run of bad luck in a short sample, which is exactly why it needs 5 races before it renders at all.

Grades start with the 2023 season. Pace, Racecraft, and Tire Management all depend on lap-by-lap telemetry, lap times, tire age, gap to the car ahead, that we pull from public timing data. That data isn't available in usable form before 2023, so earlier seasons don't have the underlying data this system needs to grade them, not a gap we've chosen to leave open.

Race control's own data sometimes can't tell a driver-caused incident from a mechanical one, most often a solo retirement or a bad qualifying session with no contact and no stewards' note attached. For those cases we manually verify the real cause against contemporary reporting before publishing the affected grade. A confirmed self-inflicted crash still counts fully against that driver's Racecraft and season Consistency; a confirmed mechanical or car-related cause is left out of the affected sub-grade rather than guessed at. A qualifying result hit by a confirmed car issue falls back to the driver's last unaffected session for the teammate comparison, while their actual grid position is left untouched, since a session fault and a genuine result are different things. Every manual override shows a note with a source link on that race's expanded grade card.

A few more specifics worth knowing. Qualifying's blend weight shifts with finishing position: a hard-fought pole leans more on the raw gap to the field, while a midfield or backmarker result leans more on the teammate comparison, since a close margin at the front is a different signal than a close margin further back. Discipline's bonus for racing hard while staying clean scales with how much wheel-to-wheel racing a driver actually did that round, so a driver cruising in clean air with few overtaking chances earns less of it than one who spent the race in traffic, even if both were equally clean. The weight each incident type carries (a causing-collision call counts for more than a procedural one, which counts for more than a track-limits warning) reflects our own judgment of severity, not a formula derived from outcomes. And every teammate-based comparison, in both Driver and Constructor Grades, is only as reliable as the pairing itself: a driver with an unsettled teammate for part of a season (a mid-season swap, like 2026's Dutch GP) has a thinner basis for those sub-grades during that stretch.

P2P Constructor Grades

Constructor grades measure team-controlled factors independent of driver talent across five dimensions: Car Pace (30%), Strategy (25%), Reliability (25%), Development (10%), and Pit Stops (10%). Car Pace blends race pace percentile, qualifying position, and speed trap data. Strategy scores three things: undercut and overcut duels between cars running close together, how a team reacts when a safety car or virtual safety car comes out, and whether its number of pit stops matched what the tyre model says was right (2025 onward). Each is weighted by how much evidence the race gave, and mechanical retirements are filtered out. Reliability classifies each retirement as mechanical or driver-caused. We research the cause of every retirement and non-start from at least two independent sources (team statements, stewards' decisions, major outlets); where we cannot confirm one, a race-control based estimate is used instead. It adds a smaller season-level signal from extra power unit elements that drew grid penalties. Reliability is scored against the rest of the field, so one team's new mechanical retirements can lift the others a little. Development uses regression analysis on pace trends with early-season dampening via sigmoid weighting. It specifically measures trajectory, improving or declining, not outright speed, so the fastest team on the grid can show a middling Development grade if their pace has plateaued, while a team that started slow and closed the gap scores higher even though they're still not the fastest car. That's by design: the grade answers “is this team getting better or worse,” not “is this team fast.” Pit Stops measures crew execution via the real stationary (jack) time from the official timing feed and lap-to-lap consistency, falling back to total pit lane time for a stop where the stationary time isn't available. The system deliberately excludes driver talent so the grade answers one question: “how well is this team being run?”

The same manually verified retirement classifications used in Driver Grades feed Reliability too, so a driver's confirmed at-fault crash doesn't drag down the team's mechanical dependability score, and a genuine car failure isn't quietly waved through as a driver mistake instead.

Constructor Grades start in 2023 for the same reason Driver Grades do: Car Pace, Strategy, and Pit Stops all need the lap-by-lap telemetry that only became publicly available that season (see above).

A team needs at least 3 graded races before we publish a grade. Strategy and Pit Stops both fall back to a neutral score for any round with no real signal to measure, a caution flag that's excluded from the season average rather than averaged in as if it meant something: Strategy when there's no genuine safety car or VSC period and every team ran the same one-stop, since there's no decision to actually grade that round; Pit Stops when we don't have usable stationary-time data for that race. And measured team-to-team Strategy differences are naturally close most rounds, so we don't force spread onto a field that genuinely performed similarly, the grade reflects real decision quality, not manufactured drama.

Pit Strategy Simulator

Pit loss, tyre wear and the undercut and overcut calculator use lap, stint and pit stop data from 2023 onward (the current cars). Green-flag pit loss is the in-lap plus out-lap minus twice the driver's own median clean lap. Under a safety car or virtual safety car every lap is slow, so those stops are measured against what the rest of the field did on the same laps instead. Tyre wear is the loss in lap time per lap of tyre age with the fuel burn added back, because lap times also fall as the fuel goes and that would otherwise hide the wear.

The calculator judges a move once both cars have stopped, so the pit loss cancels and what remains is the tyre-age difference over the laps the second car stayed out, a slow first lap on new tyres, and any traffic. The probability comes from simulating that thousands of times with each circuit's measured spread, plus an allowance for stops that go wrong.

We checked it against 515 real attempts since 2023, predicting each one without its own race. The percentages are close to what happened, undercuts are predicted better than guessing, and overcuts are not (only 17% worked). The simple rule of the gap and the rival's reaction time does about as well as the full model, so treat the tyre detail as context, not as a forecast.

All-Time Driver Rankings

Our Elo system adapts the pairwise rating approach originally designed for chess to multi-competitor races, covering every Formula 1 season from 1950 to present (75+ years, 1,000+ races). Every race, each driver is compared against every other finisher, with ratings adjusted based on expected vs actual finishing order. We era-normalize so performances from the 1960s are comparable to the 2020s. Sustained Greatness blends peak rating with consistency across a driver's best seasons, rewarding longevity at the top rather than a single exceptional year.

Season Projections

We run 50,000 simulated seasons from the current standings, playing out only the races and sprints still to come, using a 19-factor Monte Carlo engine. Inputs include driver and team base pace, circuit-specific performance history, qualifying pace deltas from telemetry, engine performance mapping by power unit, reliability rates, and recent form momentum. The result is a probability distribution for every championship outcome: not a prediction of who will win, but how likely each scenario is given current data. Probabilities update after every race as new results narrow the range of possible outcomes. Preseason assumptions about engines, regulations and rookies are deliberately trusted less as real results come in, fading out over roughly the first eight rounds.

Clinching Scenarios

Unlike the projections, this page doesn't simulate anything. It works out exactly which results would let a driver or team win the championship, or rule them out, by searching the races and sprints still to come, so every “clinched” and “eliminated” label is mathematically certain. It counts the points each event actually awards and settles ties on points the way the rules do, by most wins, then most second places, and so on. The title odds shown alongside come from the Season Projections model above.

Points Above Replacement (PAR)

PAR measures how many more championship points a driver scores compared to a replacement-level driver in the same car. We define replacement level as a driver who captures 25% of their team's total points per seat, derived from the 20th percentile of 236 team-seasons from 2000 to 2025. PAR is computed per-round and per-team, so mid-season driver swaps are handled correctly: each stint is evaluated against that team's output during the same rounds. PAR accumulates over the season like baseball's WAR: more races means more opportunity to add or lose value. A PAR of zero means the driver is performing exactly at replacement level for their machinery.

Team Principal Simulator

The Team Principal game uses real 2026 driver ratings, team budgets, and car performance data as its foundation. Driver development, contract negotiations, and race simulations are all grounded in the same statistical models that power our championship projections. Career mode runs as long as you choose, 5, 15, or unlimited seasons, with compounding decisions that reflect how real F1 teams are built over time.

Data Sources

All data on Pole to Paddock comes from publicly available F1 records and public timing data. We do not reproduce proprietary F1 content.

• Historical Records: Race results, qualifying times, driver and constructor standings, and lap data from 1950 to present, from public community-maintained archives.
• Session Timing Data: Lap-by-lap positions, pit stop timing, car telemetry, weather, and race control messages from public timing data. Used for grades, stint analysis, track domination and the H2H Compare tool.
• Steward Decisions: Penalty and licence point decisions from publicly published steward documents, used to power the Penalty Points Tracker and the Discipline grade.

Race data is ingested and processed by our statistical pipeline after each race weekend before being published on the site. Computed outputs like grades and championship projections are cached and served via CDN.