Why F1 strategic thinking must encompass more than pitstop timings

For many the 1982 Austrian Grand Prix will be remembered for the incredibly close finish as Elio de Angelis’s Lotus claimed victory from the Williams of Keke Rosberg by just 0.05 seconds.
For me, however, the outstanding memory was being in the pitlane as Nelson Piquet came in for what was the first planned pitstop for tyres and
fuel of the modern era. 

I was with Toleman at the time and although both our cars retired early, I’d stayed on the pitwall to watch the race unfold. In those days there was no pitlane speed limit and the velocity at which the car entered the pits was staggering. The stop itself took around 25 seconds, 10 times today’s standard.

It was the start of a new way of thinking about how to execute a race. Although refuelling was banned for a while in 1984, the die was set and the understanding that a grand prix did not have to be non-stop was embedded.

In 1994, when refuelling was re-introduced, teams started to think about the subject in a more mathematical way and the concept generally known as race strategy was born. In those early days the software essentially worked on minimising race time, taking into account that the less fuel was in a car, the faster it would lap – while conversely the older the tyres were, the slower the lap  would be.

These simulations were what are called deterministic. In other words, one set of conditions was imposed, and one outcome was determined. The conditions would be the effect of fuel load on lap time, the loss of lap time as the tyres degraded, and the actual stop laps and stop times. 

This was helpful but, while minimising race time was an obvious objective, a less obvious one was that track position at critical times of the race was often more important. I always say that the objective is not to complete a race in the quickest possible time but to win it in the slowest possible time.

The return of refuelling in 1994 encouraged teams to consider how different fuel loads would impact total race time

Photo by: LAT Photographic

This leads to the concept of the importance of track position and an understanding of when the driver needs to be driving at their maximum and when tyre and fuel conservation are more important. It also means that the probability of overtaking another car at any point in the race needs to be introduced.

This led to software based on a well-known statistical technique called ‘Monte Carlo’ simulation being required. The technique has nothing to do with the Monaco Grand Prix but is a technique developed in the 1940s to lead experimental work in many fields.

A Monte Carlo simulation runs many times, varying the inputs to the model within bounds set by the programmer. Each time it calculates a result and then, when the process is finished, determines which inputs are most likely to give the most favourable result. Such a simulation is called stochastic since it doesn’t provide a singular answer but instead gives a probability distribution of outcomes.

If tyre degradation is high then the undercut becomes powerful for gaining track position. However, if you try to anticipate this, your competitor may also anticipate it – so how do you deal with this?

Such a technique now leads to the possibility to expand strategy to a much more generalised meaning. For example, the software would obviously show that fast pitstops are required and hence investment in pitstop equipment will be rewarded. Perhaps slightly less obvious is investigating the effect of downforce.

Physics-based simulations will determine the best wing settings for optimum lap time but are these the best for racing? Strategy software might find that a lower-drag set-up may be slightly slower but give a better race result due to enhanced overtaking opportunities. This then means that wings with the right level of downforce and drag must be designed, tested and manufactured in time to be available at the race. This is the true meaning of strategy – a plan to achieve long-term goals.

During the race weekend the teams work in a much more tactical way. Firstly they use free practice to refine the numbers they’ve been using for tyre degradation and relative performance of their cars. The multiple simulations are then re-run to find the best approach to achieving success on race day.

Race strategies begin to take shape in practice as teams understand what their tyres are capable of

Photo by: Mark Sutton / Motorsport Images

They will also be introducing a new factor into their simulations using a technique known as game theory. This uses mathematical techniques to look at strategic interactions between the teams and how to react to the many conflicts of interest that exist in determining the best tactical approach to the race.

For example, if tyre degradation is high then the undercut, the action of making a pitstop before your competitor, becomes powerful for gaining track position. However, if you try to anticipate this, your competitor may also anticipate it – so how do you deal with this? The answer may be to stop even earlier or to do the opposite and try and conserve tyre life early on with the intention of stretching the first stint. The software will help you decide.

Of course, as you get into the race itself some of the variables become constants. Tyre degradation is now determined in real time from both your cars and those of your competitors. You may get an idea of degradation on other compounds as well.

You will also know your true position and who you’re really racing, and even see what the overtaking possibilities are. This leads to continual re-running of the simulations to give tactical awareness not just of the race as it is panning out, but also to guide what to do in the case of a virtual or physical safety car intervention. 

All of this leads to a need for powerful computation. The tens of simulations of the 1990s became tens of thousands by 2005 and millions by the end of the last decade. The next step will undoubtedly be the application of machine learning to better understand how your rivals may act and how best to react to their actions.

Advanced computing power that teams have access to helps to ensure they have all the angles covered

Photo by: Sam Bloxham / Motorsport Images