Week 1: The Pit Wall#

The words we need#

• the call — a strategic decision made during a race, on the radio, now
• to pit — to come into the pit lane and change tyres, ending one stint and starting another
• compound — one of the tyre specifications a car is allowed to run, each with its own grip and its own rate of decay
• the safety car — a neutralisation that slows and bunches the whole field, and changes what every strategy on the grid is worth

Nothing here assumes you follow the sport. Everything the course needs, the course teaches.

What a call actually looks like#

Three of them, in the form they are made in:

• pit now, or leave him out three more laps
• the harder compound and one stop, or the softer one and two
• cover the car behind when he stops, or ignore him and run your own race

Each is a choice between named alternatives, made in seconds, from incomplete information, while the thing you are deciding about is still happening.

Where it is made#

The pit wall: a row of seats on the edge of the pit lane, screens, radio, and the race running in front of you. Ours is a forty-seat room in the Faculty, and you will make your calls from it.

The wall writes everything down. Each call, the second it was made, and what your own instrument — the race simulator you spend this semester building — was recommending at that second. A departure adds the actionable limit cited and the evidence available at the call that its observable condition applied.

That log is the only reason this course can mark anything.

How the job lets you decide#

A race contains multiple consequential calls. A strategist decides when to stop, which compound goes on the car and how to answer a rival while the state of the race is still changing.

Those calls do not return separate labelled outcomes. They all contribute to the one finishing result the race eventually produces.

And what you get back#

One finishing position per race.

Multiple decisions contribute to it, along with interruption, weather, rivals and everything else that happened after each call. The result does not label which part came from the reasoning and which part came from the draw.

Each of those results is one draw#

The race you ran is one realisation of a distribution you never see.

You pit on lap 31. Two laps later a car stops at turn 4, the safety car comes out, and the race you were deciding about is no longer the race you decided in. Nothing about your reasoning changed between lap 31 and lap 33.

Run the same call against the same race again and the car does not stop. The result is your call plus everything that happened to you, added together and handed back unlabelled.

The same argument, as a picture#

So the feedback cannot carry the load#

Outcomes this compressed and this noisy are evidence — just not enough of it. The signal in any one result is small because every call and every event has been folded into the same number, so you cannot reliably tell a good call from a lucky one.

That has a second consequence, and it is the one that binds this course: feedback too noisy to learn from reliably is too noisy to grade fairly.

Two strategists can make the same call and finish a lap apart. Mark the finish and you have marked the difference between their afternoons.

What that buys, and what it costs you#

Five marked items, and none of them is a result:

The last three are one project on three dates, and the race’s ten per cent is conduct only — you cannot hide behind a good afternoon. Each page carries its own brief; read them in week 1 rather than in week 8.

What you know, and when you knew it#

A race weekend arrives in order, and the order is the constraint:

• practice — the only running you get on this track before it counts, and where what you know about the car and the tyres is bought
• qualifying — sets the grid, which is where your race starts from and who is around you
• the race — where the calls are made, and where no new information about the car is cheap any more

By the time a call is due, the measurements it rests on are hours old and were paid for earlier in the weekend. Week 3 is where that bill gets itemised.

And it has exceptions, and it is edited#

B6.3.6 is not a law of physics. Three things it does not say:

• rain suspends it. Use an intermediate or wet tyre at any point and the two-compound requirement no longer applies to you.
• the penalty is a penalty. Disqualification — unless the race is stopped and cannot be restarted, when it becomes thirty seconds.
• it changes. The FIA World Motor Sport Council mandated three sets at Monaco — ordinarily two tyre changes — and that clause was still in Article B6.3.6 in Issue 05, February 2026. It is gone from Issue 08, August 2026. Same article, same season, different decision space.

A regulator is a participant in the strategy problem. Which issue of the rulebook you modelled is part of your answer.

Twelve weeks, four acts#

• weeks 2–4 — the time trial. One car, alone. What a lap time is made of, what a stop costs and what you are allowed, and a loop that turns a strategy into a finishing time.
• weeks 5–6 — it is a contest. Other cars. Track position as a resource, and a rival who re-optimises when you do.
• weeks 7–9 — it is uncertain. Interruption and weather as hazards rather than events, and a distribution over finishing positions instead of a number.
• weeks 10–12 — it is judgeable. Whose race you are actually running, what the wall shows and what it costs to read, and whether your 60% happens 60% of the time.

Then the race, live, on a draw the whole cohort shares.

This week#

Nothing is assessed. There is no deliverable and no marker in the room.

The session is an optional drop-in for your development environment, in the pit wall, in week 1’s slot. Its page lists three lines you want true before week 2, whether or not you come.

Week 2 starts where this argument does: you cannot see what a lap time is made of, and you have to measure it anyway.