Game Theory and unusual F1 strategies (Part 2)

Following on from part 1, here I look at some more unusual F1 strategies:

1) Qualifying chaos: Italian GP, 2019

It may seem like the point of qualifying is to set the fastest lap possible, but that’s not actually the case: the point is to be as well placed for the race as possible. The 2019 Italian Grand Prix qualifying highlighted this subtle distinction beautifully.

The Q3 session was red flagged half-way through, giving the field just enough time to complete an outlap and a flying lap when the session restarted. However, there was a problem: due to the slipstream effect, at Monza it is considered an advantage to be behind another car. This meant that no-one wanted to lead the field over the line. Hülkenberg found himself at the front of the pack, and duly went off the circuit, allowing others to go through (although it has not been formally established that this was intentional). Stroll then slowed to a crawl to attempt to ensure Hülkenberg still returned to the circuit ahead. With no-one wanting to be the lead driver, everybody slowed down and time ran out before most of the field could start their fast lap. Even those that did had cold tyres and did not improve their time.

Fig 1: Radio message to Stroll, just before Hülkenberg went off track.

Monza 2019 Reactions

Opinions on the situation were mixed. Many pundits and drivers were exasperated, and vented their frustration at the lack of a qualifying finale that people were expecting. The fact that the cars crawled around at such a leisurely pace at Monza, the Temple of Speed, made the situation even worse. Charles LeClerc (the pole sitter), described it as a “mess”, whilst Toto Wolf described Hülkenberg’s actions as “not worthy of Formula 1”.

Others had different opinions, with some fans suggesting that the qualifying was interesting in its own way, or that there was no such thing as bad publicity. From my perspective, the session was a fascinating example of game theory in action. Most of the drivers acted quite rationally. What is the point of leading the pack across the line if it gives everyone an advantage except for you? It is also interesting that the unimpressed drivers in the pack did not themselves try to rectify the situation by becoming the lead driver themselves. This is obviously because they knew that to do so would put themselves at a disadvantage. From a pure game theory perspective, perhaps one would expect the driver who was due to start last out of the group (Stroll) would bolt earlier, but other than that everyone’s actions were a natural reaction to the situation.

Stroll, Hülkenberg and Sainz were given reprimands for driving too slowly, but one quote from the stewards was key:

“The Stewards strongly recommend that the FIA expedite a solution to this type of situation.”

In other words, “Don’t hate the player, hate the game.” You can’t blame the drivers too much for acting in their own interest.

2) Tyre gambles: European GP, 2007

Tyre choice in wet races can be a tricky business. A good general rule is to be on the right tyres for the current conditions. A pit-stop may cost 20-25s, but being on the wrong tyres can cost 10 seconds per lap. That’s not to mention the risk of crashing out of the race. Pitting for tyres based on the predicted weather in the future is therefore a big gamble, but one that has the potential to pay off beautifully.

Fig 2: Winkelhock passing Räikkönen for the race lead early on.

Marcus Winkelhock qualified dead last for the European Grand Prix. This was unsurprising given that it was his debut race and he was in the slowest car on the grid. However, his Spyker team took the decision to pit him for wet tyres in the parade lap, despite the track currently being bone dry. The heaven’s opened soon after the start, and literally every other car pitted at the end of the first lap (with the exception of Kimi Räikkönen, who attempted to pit before understeering back onto the circuit). Winkelhock then proceeded to overtake Räikkönen on the track and pull away from the rest of the field in a rare David vs Goliath moment for the sport.

Fig 3: Several drivers went off in the early rain.

Red Flag

With several cars spinning off at turn 1, including renowned wet weather drivers such as Lewis Hamilton and Jenson Button, the safety car was swiftly deployed. It was quickly followed by a red flag. Spyker then found themselves leading the race due to their audacious gamble. They proceeded to gamble again, fitting full wet tyres on at the restart when everyone else had intermediates. This roll of the dice was an unmitigated disaster, with the field quickly passing Winkelhock.

So, what can we learn from all this? It’s a classic example of risk-reward analysis. Expectations were low for the team going into the race, and the worst case scenario of their early gamble was that they’d be half a minute down on the rest of the field. This is not a huge punishment given that they’d expect to be running at the back anyway, and any incoming rain/safety cars would have allowed them to close the gap. With little to lose they rolled the dice and lucked their way to the front (although Winkelhock deserves praise too for his handling of the situation).

Risk-Reward

The issue is that after the red flag their risk-reward analysis was now askew. Of course they still had an inexperienced driver and a poor car, so the expectation was that the field would quickly make their way through, but there are 2 major reasons why their strategy was a mistake. Firstly, although the best cars would have overtaken quickly, it would have taken significantly longer for the rest of the field to do likewise. It seems bizarre to gamble away their winning position so freely.

The second issue is that almost a quarter of the field had spun out of the race on lap 2, and it was still a wet race. This makes the possibility of hanging onto a points position significantly more likely, a reward that would have been richly deserved. A conservative approach would have been much more logical the second time around. Their tactics almost feels like a gambler so convinced they’re on a winning streak that they abandon all reason and keep going for broke until it blows up in their face.

Ultimately it proved academic as Winkelhock’s car broke down after just 13 laps. Tragically this would be his only F1 start. Just 2 years later Ferrari tried something similar with Kimi Räikkönen at the 2009 Malaysian Grand Prix, only for him to trundle round on full wet tyres(!) despite the track still being dry. Gambles do not often pay off, but when they do, the 2007 European Grand Prix suggests that one should not push their luck.

3) Racing under protest: Japanese GP, 1997

I mentioned this one briefly in my previous post on unconventional strategies, but it is worth looking into in more detail. Going into the weekend, Jacques Villeneuve held a 9 point lead over Michael Schumacher in the championship with 20 points still on the table. Villeneuve qualified on pole, but several drivers, including both Villeneuve and Schumacher, were punished for ignoring yellow flags. Unfortunately for Jacques, he was already on his final warning, with the punishment being a race ban. It may seem like an extreme punishment for a comparatively minor offence, but those were the rules at the time.

The bigger question is not why the punishment was so severe, but how on Earth Villeneuve and the Williams team had found themselves caught out by yellow flags again, knowing full well what the punishment would be. Their risk-reward analysis was clearly out of kilter. Williams protested the punishment, and he was allowed to race, but the expectation was that the ban would still hold and he would subsequently be disqualified. There were even suggestions that the protest was a mistake, and that the FIA would enact further punishment as a consequence of it (such outcomes were known to occur at the time).

Race Strategy

For race day, both Williams and Ferrari faced a strategic minefield. Villeneuve’s main aim was not to win the race, as he would likely be disqualified anyway, but to do his best to prevent Schumacher scoring strongly. Ferrari meanwhile, had to somehow win the race despite their main rival actively sabotaging them.

Jacques lead away from pole, and proceeded to drive at a glacial pace. His strategy was focused on getting other drivers to pass Schumacher, either directly or through pit stop strategy. An attempted overtake by Schumacher would be risky; if they happened to tangle, Villeneuve would lose nothing. Schumacher’s team mate Eddie Irvine then went around the outside of Mika Häkkinen and Michael Schumacher at one of the S-curves. Given that this is not an overtaking spot, something was amiss. It turns out that Schumacher was also playing the strategic game, and he had engineered a situation where Irvine was able to go through. Irvine then muscled his way past Villeneuve and pulled out a 12 second lead in 2 laps. Williams began to urge Jacques to push on, perhaps with one eye on the constructors title.

Field Spread

Eventually he obliged, and the field spread out. Schumacher then passed Villeneuve during the pits stop window, with the Williams driver aggressively squeezing Schumacher when he emerged from the pits. (Schumacher lambasted Villeneuve for driving in his manner. A suggestion that become highly ironic given that he intentionally crashed into Villeneuve at the following race.) Irvine then slowed to let Schumacher through, before giving Jacque a taste of his own medicine; blocking him so that Schumacher could get away.

Fig 4: An ecstatic Schumacher celebrating victory

In the end Schumacher won the race, whereas Villeneuve could manage no better than 5th. Williams eventually dropped their protest and accepted the disqualification, which set up the infamous title decider. So, what did each team do well, and how could their strategies have been improved?

The good
Williams: The protest.

Villeneuve had been disqualified from the race before it even started, and whilst his result ultimately didn’t count, he was at least allowed to race. Whilst Jacque was ultimately unsuccessful in his attempt to sabotage Schumacher’s chances, his actions did cause Ferrari to run a fundamentally sub-optimal strategy, which contributed to Williams securing the constructors title that weekend.

Ferrari: Using Irvine.

The Ferrari strategy that unfolded seems almost laughably complex in hindsight. Firstly, Irvine got himself into the mix at the front. Secondly, he forced his way past Villeneuve. Thirdly, Schumacher forced his way past Villeneuve and lastly Irvine then let Schumacher through whilst blocking Villeneuve. Of course there are other ways it could have gone down. Their job would have been made significantly easier if Schumacher had passed Villeneuve at the start, for instance. Nevertheless, their plan of fighting Villeneuve’s ungentlemanly tactics with similar tactics of their own was incredibly effective at delivering their goal- a Schumacher victory.

The bad
Williams: Villeneuve’s mixed strategy.

Jacque Villeneuve had a plan to hold up Schumacher… a plan that he abandoned before the first round of pit stops. Perhaps in hindsight it would have been better to continue the strategy for longer, thereby giving other drivers/teams the opportunity to leapfrog both of them by going short/long. Schumacher was Villeneuve’s target for the race. and yet he seemed to get distracted and chase after Irvine before his tactics had borne fruit.

Williams: Going solo.

Whilst the Ferrari drivers teamed up, Williams seemingly never considered doing the same with their other driver Frentzen. Arguably this allowed Frentzen to run his own race, which duly delivered the 2nd place the team needed to win the constructors championship. Was this the ultimate goal, or did the team miss a trick? If Villeneuve’s strategy was to hold up Schumacher, it should have been relatively simple to engineer a situation where Frentzen could leapfrog both of them. This could be attempted on the track (as Irvine managed to overtake Häkkinen and Schumacher with the letters assistance) or in the pits (Frentzen could have been fuelled long, allowing him to put in fast laps on low fuel after Schumacher pitted, for example).

A Frentzen victory would deliver the team the constructors title whilst also helping Villeneuve in the fight for the drivers title. If Irvine ended up leading the race ahead of Frentzen, then Ferrari’s job would be even more complex, given that Schumacher would now be down to 3rd place (or 4th if Villeneuve could stay ahead of him too).

Ferrari: Irvine’s 3rd place

As alluded to earlier, Irvine’s strategy was sub-optimal. Not only was he forced into a 3 stop strategy to allow him to be competitive in the early stages, but in the mid-stages of the race his role changed to blocking Villeneuve, which ultimately cost him second place. Given that his put Ferrari out of the running of the constructors, it could be argued that Ferrari were too focused on the driver’s championship.

As referenced in part one, conditions for a “win” in game theory can vary significantly for different players, and perhaps the actions and strategies are explained by Williams having different priorities to Ferrari (constructors championship vs drivers championship respectively).

2 thoughts on “Game Theory and unusual F1 strategies (Part 2)

  1. Pingback: Contents Page – Motorsport Analysis

  2. Pingback: Game Theory and unusual F1 strategies (Part 1) - F1 Analysis

Leave a Reply