Skip to main content

Week 8: Weather

Compare irreversible wet-weather commitments across a supplied forecast ensemble, preserve the position-first objective, and state the downside

26 April 2027

Open the slides

Last week one interruption might arrive before one bounded call. This week the uncertainty has several shapes. A forecast path is one possible sequence of future weather states. A forecast ensemble is a set of those paths, available before we know which weather will occur. It does not identify one certain future.

Our output is an irreversible wet-weather commitment for a supplied synthetic scenario. A wet-weather set is one physical set of tyres supplied for the conditions this fixture labels wet. A commitment rule names the lap after which the strategy switches to that set. We make the rule irreversible inside this exercise so that hindsight cannot rewrite it. Real teams are not forbidden from revising strategy; the restriction belongs to our fixture.

We first validate the synthetic path weights: every weight must be finite, non-negative and together they must sum to one. We also validate the legal candidate laps and the physical-set ledger. If the required set has been returned, the dependent commitments are infeasible and never reach the scorer. For each remaining candidate, we score every supplied path and calculate its weighted mean finishing position. Lower position wins. Weighted mean adjusted finish breaks a tie in position and does nothing before one.

Then we record the chosen lap, its score and its reason before the realised path is revealed. The reveal tells us what happened; it does not change what the forecast said when the call was due. We also report the selected rule’s downside: the path or paths producing its worst finishing position, and whether rain arrived earlier, later or not within the supplied horizon. That is a separate report, not a minimax objective.

ECMWF’s ensemble work supports the real idea that several forecast realisations represent uncertainty and add information beyond one best guess. It does not supply this race. Every path, weight, outcome, tyre effect and commitment restriction here is synthetic, and the weights are not calibrated.

The lab saves one auditable rule and its downside. The final instrument must carry the same forecast-conditioned commitment, the assumptions beneath it and evidence that the call existed before the outcome.

Teaching team