// Pit Wall

You are the strategist.

Pick a track and a grid slot, then race 19 scripted rivals over 57 simulated laps: same safety cars for everyone. A reference car (Medium → Hard, pit lap 24) has your pace and your luck, so the gap to it shows what your strategy alone is worth. Decide when and what to fit, and see where you finish. Built on the fitted numbers of my F1 strategy study, in a stylised model world: a game, not a prediction of real races.

What this is, and where the numbers come from

A stylised model world, not a prediction of real races. Generic cars and three tyre compounds. No traffic, no virtual safety cars, no other drivers.
From the study (timing data, 2022–2024): pit-stop loss 22.99 s with a core spread of 1.37 s and a 6.3% slow-stop tail; lap-time noise as AR(1) with φ = 0.42; safety-car hazard of 0.91 per 100 laps after lap 3, with a higher lap 1–3 spike. The loss for a stop under a safety car is only weakly measured (109 of 453 stops had a usable value), so treat “cheap stops under a safety car” as indicative.
Assumed, because the data could not identify them: the compound offsets, the tyre degradation rates, the “cliff” past a compound’s usable life, and the fuel-burn prior. All of these have sliders: the compound offsets and the fuel prior each have their own, and one slider scales degradation and the cliff together. Two more sliders (noise, safety-car frequency) test the study’s own numbers. Tyre sets per compound are a race rule you choose, not a fitted number.

Tracks and rivals are generic. The four tracks are illustrative: pit loss differs a lot between real circuits in the study (roughly 18 s to 29 s), but the lap counts, base pace, wear and incident rates here are assumed. Rivals are scripted cars (a fixed pace by grid slot, a one- or two-stop plan, and a stop under a safety car when theirs is due soon) with no overtaking or traffic, so places are decided by total time. Your finishing place is a result in this game world. The study did not validate finishing-position predictions and makes no claim about them.
Points: the top ten score 25, 18, 15, 12, 10, 8, 6, 4, 2 and 1 (a familiar scale borrowed as a scoring rule), and the tally runs across the races you finish in one visit. Nothing is saved: it resets when you reload the page.

Project report Repository Unofficial project, not affiliated with Formula 1.

Set up the world

Track: Start from:

The study’s typical race: average wear and incident rate. 57 laps, pit loss about 23 s.

Start tyre:World #

Tyre sets set aside for the race, per compound: The set you start on counts. A compound with no set left can’t be fitted.

Assumptions you can change (the study could not pin these down)

Tyre degradation multiplier: 1.0× Lap-time noise inflation (variance): 1.0× Safety-car frequency: 1.0× Soft pace vs Medium: −0.50 s/lap Hard pace vs Medium: +0.45 s/lap Fuel-burn gain (time saved each lap as the car gets lighter): 0.035 s/lap

Compound offsets, tyre rates, the tyre “cliff” and the fuel-burn prior are my assumptions: public lap data cannot identify them (between-race variation was too large). The first slider scales tyre rates and the cliff together. Pit loss (23.0 s on the study-default track, spread 1.4 s, 6% slow stops), noise (AR(1) φ 0.42) and safety-car odds (0.9 per 100 laps, lap 1–3 spike) come from the study.

Lap

0 / 57

Position

P10 / 20

Your tyre

–

Tyre age

–

Vs reference

0.0 s

Pit this lap:

Stopping costs about 23 s. A dry race needs two different compounds (30 s penalty otherwise). Safety cars arrive unannounced, and they make a stop cheap.

Standings

Gap to reference car, lap by lap (below the line = you are ahead)

Your place, lap by lap

Table view

What this is and isn’t. Finishing places are outcomes of this game world, not predictions. The reference car and your car share all noise, so only the strategy differs. Your replay keeps your pit laps fixed, so it does not capture reacting to safety cars as you did live. Virtual safety cars, traffic and overtaking are not modelled; rivals are simple scripted cars and tyre rates are assumed. The “break-even multiplier” is the tyre-wear level at which your plan and the reference would tie on average. Real strategy decisions would need the data and validation described in the project report.

// Work with me

Have a project in mind? Let’s talk.

Website, dashboard or full product — I’ll take it from the first conversation through to launch and maintenance.

mikailadham04@gmail.com +60 16-208 1721 github.com/mikailadham

You are the strategist.

Set up the world

Standings

Gap to reference car, lap by lap (below the line = you are ahead)

Your place, lap by lap

Result

Your time against the reference car

Where you finish

Have a project in mind? Let’s talk.