2021 Season pit stops: Rounds 1 to 7

The pit stops of the season

The Championship battle seems to be shaping up to be a close one this year. With such small margins, the pit crews may end up being crucial in the World Championship decider. Let’s take a look at the pit stop crew and their performance in first 6 races of the year

Methodology

The pit stop data was obtained from DHL’s Fastest Pit Stop Award website.  This data was then compiled and processed in a few distinct ways.

For the full data analysis, the data was used without filtering any data points. The information was grouped by either driver or team and then the average time and other descriptive statistics were calculated from ALL the points.

As for the filtered data analysis, the data was grouped by either driver or team and filtered using the Generalized Extreme Studentized Deviate test (GESD). This test progressively evaluates anomalies, removing the worst offenders and recalculating the test statistic and critical value. It has shown better performance than IQR filtering methods in some previous scenarios so I decided to use it for this analysis. One important note about this method is that I limited the number of anomalies to a maximum of 20% of the data. This way if a pit crew has been making many mistakes, some of them will remain on the data and the team/driver will be penalized for them. 

Finally, for the pooled data analysis, the data was grouped by either driver or team and then filtered using the GESD method.

After the data was processed, I created some nice charts for you to see. I hope you enjoy them.

Interpretation

For the non-filtered data, you will see that outliers—abnormally low or high values— will drag the mean (average) time drastically towards the left or right of the plot. This average time will be arithmetically sound, but perhaps not representative of the average pit stop time you will see from a particular pit crew. It is because of this that the filtered data was used, to create a more representative average pit stop.

The charts have three distinct intervals which represent the density of the data. First, the 50% interval contains 50% of the data and it means that you would expect most pit stops to be around that time. Second, the 80% interval is wider and contains perhaps some pit stops that were abnormally fast or slow. Finally, the 95% interval contains most of the data, including perhaps abnormally fast or slow stops. If a data point is not included in any of these intervals, it can be considered as a bit of an outlier.

As you will see, these intervals are asymmetrical for some drivers/teams. These intervals tend to be shorter on the left side, and longer on the right side. This is because it’s easier to make a mistake and get a slow pit stop, than being perfect and getting a quick pit stop.

One final note. A couple of years ago I used the geometric mean for my analysis instead of the arithmetic mean. This year I decided to go with the arithmetic mean since it’s a lot easier to interpret. While it has the downside of being massively influenced by outliers, it shouldn’t affect the filtered analyses since the outliers were already removed.

Team pit stop times

To see the hi-res png images click here and here.

Full data

Let’s start with the teams, shall we? First, we will take a quick look at the full data chart, and then proceed to properly analyze the filtered data plot.

In the non-filtered department, Ferrari lost their 1st place after having a bad weekend at the Circuit Paul Ricard. Williams took the first position after an error-free weekend in France. The Alfa-Romeo crew continues to be the slowest one, but only due to the massive influence of a 35+ seconds pit stop. At this rhythm, they will overtake the Haas crew soon as long as they avoid another very long stop.

As said in the previous post, these numbers are not very representative of the performance of the pit crews. Outliers have a strong influence on the average time, skewing the results and our perception of the pit crew’s abilities.

Filtered data

The filtered data shows a different picture, with Red Bull leading with an average time of 2.33 seconds per pit stop. Williams now trails Red Bull in second place with an average time of 2.44 seconds.

Mercedes has been criticized for their performance at the pits, but they have climbed the charts to 3rd place after 7 rounds. While they haven’t been as consistent as Williams, they have had very quick stops and sit deservedly in third place at the moment.

Ferrari was the biggest loser after having a bad weekend. Their average time went from 2.63 to 2.74 seconds per stop. Their fastest stop at the 2021 French GP was of 2.73 seconds, which is not good enough for a team of Ferrari’s quality.

McLaren had a good weekend, but their filtered average increased. Why is that? Last week the GESD algorithm detected a 3.93 seconds stop as an outlier, and removed it from the data. This week, the same algorithm didn’t detect that point as an anomaly. This is normal, and while it may seem unfair right now, it works the same for all teams. If McLaren continues to do a good job, the algorithm will eventually remove that data point from the analysis and their average time will improve.

Driver pit stop times

To see the hi-res png images click here and here.

Filtered data

Max Verstappen’s crew continues to lead, albeit by a smaller margin over Kimi Raikkonen’s crew. Verstappen’s times went from an average of 2.19 to 2.24 seconds per stop after an average weekend at the 2021 French GP.

Kimi’s crew had a poor weekend, so poor that the system detected that stop as an outlier. As stated before, the outlier detection algorithm limits the number of outliers to only 20% of the stops. Why is this relevant? If a pit crew continuously makes mistakes, eventually the algorithm will leave some of those slow stops on the data, increasing the average pit stop time.

The biggest winner this weekend was Sergio Perez and his crew. With a fast pit stop of only 2.04 seconds, Perez’s crew climbed several positions and now ranks as the 4th fastest of the season.

Currently, the pit crew of 14 drivers has recorded an average time of less than 3 seconds. Even more impressive is the fact that 17 out of the 20 pit crews are separated by less than 1 second on average.

Pooled pit stop times

To get a hi-res png image click here.

The filtered pooled pit stop data is just meant to visualize the pit stops done in this season as a whole. In this chart, the top stacked histogram shows how the times were distributed with an interval of 0.1 seconds. For example, all 100% of the pit stops done between 1.9 and 2 seconds were done by Red Bull. For times between 2 and 2.1 seconds, 75% of them were done by Red Bull and 25% by Aston Martin.

The average times shown at the filtered team data are also shown in this plot. You can see how Red Bull leads this way, but most teams have respectable averages. The only pit crew that stands out is Haas’ with a poor time of 3.60 seconds per stop.

Final remarks

First of all, I hope you have enjoyed this analysis. I am intending to make this another one of my weekly/biweekly series. I think that this is a type of analysis that can easily be updated every week to keep you up with the performance of the pit crews.

Red Bull continues to dominate, but Williams is not that far off behind. Mercedes’ crew took 3rd place after the 7th round of the season, and if they manage to stay consistent, they may end up playing a crucial role at the end of the season.

Finally,  I wanted to let you know that working on articles like this one is no easy task. If you enjoy visiting my site, please consider sharing my posts on social media or with friends and family. If you want to support me with a donation, just click on the about tab on the menu and you’ll find some options to help me out.