All articles

How this is calculated, and where it fails

A simulator is only worth anything if you can argue with it, and for that you have to show how it decides. These articles explain what the numbers on this site are made of: what weighs how much in a race, where each driver and car rating comes from, and how they are corrected against real results. What does not work is here too. There are five seasons the model cannot explain, and they are named rather than hidden. Starting first is worth half the race, and the circuit matters far less than it looks.

5 articles

MethodPredictionsSimulator

Our simulator's scoreboard: four races called before they ran, and the grid still beats us on average

Since July we publish each Grand Prix preview before the race and the result after it, recalculating nothing. Four races in: 2.49 places of average error, 30 of 32 top-8 calls and 32 of 40 points scorers. Plus an awkward number we are not hiding: assuming they finish in starting order misses by less (2.25). Where we do win is Monza, when the grid lies.

22 September 2026 · 7 min read

MethodReliabilityRetirements

We dropped a car’s reliability from 99 to 60 and it finishes in the same place

Every team has a number called reliability. We covered two thirds of its scale on the 2026 grid, 60 championships per step and identical seeds, and the front-running team loses 0.28 championship places. The consequence shows in our predictions: the model expected 4% retirements across the four Grands Prix it called in advance, and the real figure was 18.2%.

22 September 2026 · 8 min read

DataCircuitsHistory

How much is pole worth? 709 races say half, and that the circuit hardly matters

Pole converts into a win 50% of the time since 1988. The surprise is the other half: a track’s reputation predicts nothing. Monza converts more poles than the Hungaroring, and the correlation between how easy overtaking is and what pole is worth is close to zero. Our own simulator gets this wrong, and we say how.

1 September 2026 · 8 min read

MethodologyHistoryVilleneuve

The five seasons our simulator cannot explain

We re-ran 30 real seasons, 100 times each. The engine picks the right champion in 25 of them. These are the five it gets wrong, and three turned on a single afternoon no simulation can see coming.

26 August 2026 · 8 min read

Simulation engineHow it worksFormula 1

How a Formula 1 simulator works under the hood

From a driver's DNA to a Grand Prix result: how a simulation engine turns attributes, car, weather and strategy into a full race. The logic behind F1 Legend Sim, explained step by step.

24 July 2026 · 9 min read