Executive Summary:
We examined the timing data from the 2019 and 2020 24 Hours of Le Mans to understand how the race changed from one year to the next, and to understand the differences between the winning teams and their competitors. Here is what we found out about the 2020 race:
- The track was slower in 2020 than 2019, but the winners of all classes completed more laps because there were less Slow Zones and Safety Cars in 2020
- In the amateur classes, LMGTE AM and LMP2, the bronze and silver drivers made the biggest impact in overall team results, since they were highly varied in their performances compared to the gold and platinum drivers.
- In 2020 the amateur classes, LMGTE AM and LMP2 were more competitive than in 2019, as the top 5 cars were more consistent in both classes. LMP2 in particular had a standard deviation that was .5s per lap smaller than in 2019, which is significant.
- In 2020, the professional classes, LMP1 and LMGTE PRO, were slightly less competitive. The standard deviation of the lap time of the top 5 cars was marginally smaller, and there were less cars overall.
- None of the class winners were the absolute best in both pace and consistency. The winners had some of both, but they also had clean races with good luck and no incidents.
Table of Contents:
- Introduction
- What is a Competitive Field?
- Quick Stats about 2020 vs 2019 24 Hours of Le Mans
- The Four Classes at Le Mans
- Amateurs vs Professionals in LMGTE
- Consistency Differences
- Pure Pace vs Results in LMP2
- Consistency and Speed Scatterplots
- Verdict 2019 vs 2020 Le Mans
Introduction:
The 24 Hours of Le Mans is one of the most challenging, enduring, and well-known races across the globe. To win the 24 Hours of Le Mans a team needs to execute everything to perfection. Rarely does anyone finish Le Mans without any problems, but the winners always have a few things in common: fast pace, few mistakes, and good luck. The 88th running of the 24 Hours of Le Mans was held from September 19-20, 2020, different from the usual June date for the race because of COVID-19. There were three more hours of night-time than usual for the race since it was in September, but the weather was mild. This year the overall winners, Toyota LMP1 #8 completed 387 laps, more than five ahead of any other team.
The WEC uses magnetic timing loops and beacons on each car to record the results of the race (,link to official timing database). This publicly available data is simply lap and sector times for each car and driver, but there is so much that can be learned about the race from this data in context of the results of the event. We will apply statistical analysis methods to the WEC data to have a first-hand look at the speed of each race car and driver, and then we will plot our results to visually to compare patterns and trends in the race. In Part 1 of our series about Le Mans, the main question to be addressed is: How competitive was the 2020 24 Hours of Le Mans and was it more or less difficult to win than 2019?
What is a Competitive Field?:
In racing it is common to speak of how ‘competitive’ a group of drivers and teams are. This refers in general to how difficult it is to win a race, but more specifically it is talking about how many cars are fast compared to the total cars in the field. A car’s ‘pure pace’ is simply how fast its lap times were throughout the race. A car could finish poorly but still have good pure pace because there are many other factors besides a car’s pure pace that contribute to its race results. For a car to have good pure pace, it needs both consistency and low lap times. We can look at the standard deviation of lap times to see how consistent the cars were, and then we can look at the average value of their lap times to see if they were fast. It’s hard to know if anyone could have gone faster than the fastest car in the race, so we use the percent difference from the fastest car in each class to find out how other teams compared to the best.
Once we have understood the pure pace of the cars in the race then we can compare the finishing results to see how closely they matched the results of which cars were mathematically the fastest in pure pace. What makes endurance racing so challenging and so unpredictable is the fact that so many external factors play a role in determining who actually crosses the start finish line first after 24 hours have passed. Though our data isn’t able to tell us about all of the mistakes, incidents, and mechanical failures that can befall a team over 24 hours, it does tell us about what is happening to the cars when they are pushing as hard as they can.
Quick Stats about 2020 vs 2019 24 Hours of Le Mans:
The 2020 edition of the 24 Hours of Le Mans changed a lot from 2019, and in some ways significantly. The event took place in September which meant that there were three more hours of darkness during the race than in 2019, but its hard to say if that made a significant change to the speed of the race since other atmospheric conditions can contribute as well. The makeup of classes changed between 2019 and 2020, which can significantly change the amount of traffic the fast cars have to deal with. This chart shows the general differences between the 2019 and 2020 event.
Table I – Comparison of the 2019 and 2020 24 Hours of Le Mans with average consistency and lap time metrics for all cars in each class.
There were significant changes to the layout of the field in 2020. The overall trend is that there were fewer professional entries in the LMP1 and LMGTE Pro classes, and more entries in the classes with amateur drivers, LMP2 and LMGTE Am. Most notably, there were nine less LMGTE Pro cars and, but the standard deviation of lap times for the top five cars remained almost exactly the same. There was a significant change in the LMP2 field in 2020. The top 5 finishers were much more consistent this year than last year. Since the LMP2 cars cannot be changed from one year to the next, it is likely that the top 5 teams performed better than in 2019. All of the classes completed a few more laps in 2020, but they also all had slower fast lap averages than 2019. Since every classes’ fastest laps were slower by at least second, that change is probably due to a minor change in track layout or different atmospheric conditions. One might expect the winners to complete less laps if their average lap times were slower, but having just a few more or less Safety Cars through the duration of the race can make a much larger difference.
The Four Classes at Le Mans:
Fig. 1 – Average standard deviation for all cars in each class, filtered for yellow and pit laps.
Having four classes each racing on track at once is what makes Le Mans hectic (visit the WEC website for official class explanation). Each class is unique as the driver experience requirements are different. It is clear that the professional classes with more experienced drivers, LMP1 and LMGTE Pro, are more consistent than the amateur classes. That alone doesn’t describe how hard it is to win each class though, since not each class has the same number of cars. It could be that the bottom 50% of cars in each class are much less consistent than the top 50% when there is a larger variety of teams, and that is what makes the difference in the standard deviation for the whole class put together. Looking at each class as a whole, we can see how the differing driver requirements affect on track performance. On a first glance, the drivers of the LMGTE Pro class were the most consistent. We also have to factor in that between the classes, some cars might be naturally less consistent than others since every type of car’s tires wear differently. Some may last longer than others and make the car more consistent. That doesn’t necessarily take away from the drivers though, as they still have to work hard to take advantage of the good tire wear!
Amateurs vs Professionals in LMGTE:
The major difference between the amateur (LMGTE AM, LMP2) and the professional classes (LMP1 and LMGTE Pro) is that the amateur classes are required to have bronze and or silver rated drivers (WEC official driver classification rules). Bronze and silver drivers are a lot less experienced than their professional gold and platinum counterparts. Including bronze and silver drivers in the race means that a lot of less experienced, older drivers, and unproven, young drivers get their chance in the race. This is a fascinating part about the 24 Hours of Le Mans that heavily influences the results and consistency of every class as we have seen. In the amateur classes it is important for all three drivers to work together to help the bronze and silver drivers be as fast as possible. A good result often can hinge on the speed of these less experienced drivers! The effect of the driver ratings on the speed of each car can be seen when we plot out how lap times were distributed for each driver over the race.
On these violin graphs, we have the lap times for the top five fastest LMGTE Pro and LMGTE Am teams over the whole 24 hours of racing. These graphs can visually tell us a lot about each team’s pure pace, and make it easy to distinguish which drivers are consistent, fast, or both. The lower the violin, the faster the driver is. The less spread out it is top to bottom, the more consistent the driver is. Best of all, is a violin that has most of its area towards the bottom where fastest lap times live.
Fig. 2 – Violin graph plotting the top five cars in the LMGTE Am class by driver.
It is easy to spot which drivers are bronze drivers in this graph. Their fastest lap times are often slower than the slowest lap times of the pro drivers. Generally, the faster pro driver’s lap time distributions are fairly similar, whereas the amateur drivers have a larger variety in lap times. That is why having a good bronze driver is a key to the race because their good performance is much more impactful than a good performance from a professional, who is already close to the limit of what is possible with the car.
Here is what the violin graph looks like for all LMGTE Pro categories. It is incredible how similar in lap times all of these drivers are, which matches how competitive the LMGTE Pro field looked on the consistency bar graph. These drivers are all professionals and all of their cars were in a narrow band of lap times.
Fig. 3 – Violin graph plotting the top five cars in the LMGTE Pro class broken down by driver.
A good example of a very fast driver is the blue bubble on car #97. This car won the race, and you can see clearly how the blue bubble has less area higher up on the chart compared to the blue bubble on car #95 that finished the race in third place. Whoever has the most area of the bubble lowest on the graph was the fastest driver, but once again that may not have anything to do with finishing order. In terms of lap time, we can’t conclude anything more than which drivers were fast and slow. To say that the fast drivers completely outperformed the slower drivers might not be completely correct since there are a variety of reasons why someone might be slower than another. It could be that the fastest driver drove during ‘happy hour’, which is the early morning hours where the hot sun warms the track but the cold air is still dense, or it could be that the fastest driver had the most brand new sets of tires in the race. Consistency can tell us more than lap time about whether or not a driver performed well, since a few extremely fast lap times among the whole driving stint don’t mean as much frequent and solid good lap times.
Consistency Differences:
Fig. 4 – The standard deviation for the top half of cars in each class by finishing order.
Fig. 5 – The standard deviation for the bottom half of cars in each class by finishing order.
Looking back at the overview of each class in the race, we split up the bar graphs to show standard deviations split between the top and bottom halves of the class by finishing order. We were curious to see how much more consistent the cars that finished in the top of their classes were than those that finished in the back half of the field. The difference in consistency was greatest with the amateur classes as expected since they have the largest variety of driver skill levels. The LMGTE Am field, of 20 cars, had the biggest change in standard deviation with the average top 50% having 3.5s average, and the bottom 50% have a 5.4s average. The LMP2 top 50% had an average standard deviation of 3.1 seconds, but the bottom 50% had an average of 4.4 seconds. Looking at this graph, it is clear the LMGTE Pro has the tightest grouping of lap times of any class. The top 50% of their class has a standard deviation of 1.8 seconds and the bottom 50% is only .4 seconds different at 2.2. That is an impressive spread which tells us that even the cars who finished toward the back of the LMGTE PRO class were consistent and had a good chance to win based on pure pace. In the amateur classes the lower half of the finishers were not consistent enough to have a chance at the win. Now we will transition to the prototype classes to see how well their results matched the speed and consistency of each car.
Pure Pace vs Results in LMP2:
Table II – 2020 race results with standard deviation and average best 10 laps per car.
Looking at the LMP2 class starting at car #22 we can clearly see how pure pace affected the race. This chart contains the finishing order of each car, the standard deviation of their lap times, and their average lap time of the top 30 laps. The trend is clear: the top five all have standard deviation of less than 2.5s, but the car that finished in second was one second slower per lap on average over the top 30 laps! It’s not a car’s best lap times that make a difference in the result, but rather how they remain consistent over a whole stint. Speed is the key to pure pace, but the standard deviation and average best laps tell us that more frequent good laps makes a bigger difference than a higher average 30 laps. It is clear that a certain team can be slower than the others in terms of absolute fastest lap times, yet they can still outperform faster cars if they are more consistent and mistake free. These numbers do leave out part of the each team’s story through the race because many other things can slow cars down that aren’t visible on the lap time records. Reliability issues, contact with other cars, penalties, and much more can reduce a team’s race effort to win to a poor result.
Consistency and Speed Scatterplots:
Fig. 6 – LMP1 average 10 best laps vs. lap time standard deviation.
Fig. 7 – LMP2 average 10 best laps vs. lap time standard deviation.
Fig. 8 – LMGTE AM average 10 best laps vs. lap time standard deviation.
Fig. 9 – LMGTE Pro average 10 best laps vs. lap time standard deviation.
Verdict – 2019 vs 2020:
What is fascinating about these scatter plots is that each confirms that the cars were slower in 2020. The orange dots for 2019 tend to be lower, which means faster, than the blue dots for 2020. The biggest takeaway though from all four of these charts is that without a doubt, each class was more consistent in 2020. The further to the left the dots are indicates that they have a smaller standard deviation of lap times, and in every plot the blue dots are generally further to the left. With this information we can come to a conclusion, which is that 2020 was more competitive than 2019. Consistency is a better judge than speed is for performance since so many factors can take away or give absolute lap time, but many less factors can take away a driver’s ability to deliver similar lap times. The difference between years is not huge, maybe the standard deviation only varied by about a half a second, but in racing a half a second is what makes the difference between winners and losers.
Drivers cannot fake consistency. The best way to be consistent is to push as hard as possible and to find the upper limit of what the car is capable of, which is what all race cars encourage drivers to do. Slowing down to drive more consistent can lead to mistakes from a lack of focus. Drivers are always learning, improving, and innovating. It is good to see that the level of performance in Le Mans was set higher in 2020. Hopefully that bar will continue to be set higher in years to come.
