2021 Pre-season testing – Part 1

Formula 1 is back! Let’s take a quick look on how the 2021 pre-season testing weekend developed.

All the laps!

The chart above shows a histogram of all the laps done during the 2021 pre-season testing weekend. The chart includes every single lap, including laps in which the drivers just took the car out for a single-lap stint. The drivers have been ordered alphabetically by team, so the order of them, in this case, means, well, nothing.

The little chart at the bottom represents the percentage of laps done by each team (so…both drivers) in each lap time bin. What do I mean by time bin? It just means that lap time was represented as a discrete variable instead of a continuous one. The bin on the left side of the plot represents all the laps done in the 1:29.000 to 1:30.000 second range. The colours represent the percentage of laps done by each tine in that particular time bin. This was done for all the time bins from 1:29.000 to 2:30.000.

Standardized histogram

We can take a better look at the same funny chart with a bigger representation of it. Take a look at how teams like AlphaTauri, Alpine and Red Bull are well represented on the fastest end of the spectrum, while teams like Haas didn’t even figure in the bins containing lap times of less than 1:32.000 minutes.

To be fair, the right side of the spectrum doesn’t show you much. I mean, are you interested in knowing how were the teams represented in lap times of, let’s say, 2 minutes or more?

Stacked histogram

The main issue with the standardized histogram is that it is, well, standardized. Making all the bins go from 0 to 100% doesn’t tell you in which lap time range did the teams spent most of their lap times during the weekend. Fortunately, we can fix this with a stacked histogram. With the stacked histogram we can see that most of the laps done during the weekend were done between the range of 1:36.000 minutes per lap to 1:39.000 minutes per lap.

Mercedes spent most of their weekend on the 1:36.000 to 1:37.000 minutes per lap range, not doing as many fast laps as teams as Red Bull did. Ferrari, on the other hand, did more or less an equal number of laps in the range of 1:36.000 to 1:39.000 minutes per lap range.

An interesting pattern was also seen in Alpine—formerly Renault— and McLaren, with both teams spending an important part of their weekend working on the fastest side of the spectrum.

Haas looks to be the slowest team as of right now. I know that people say that you shouldn’t conclude from pre-season times, but I disagree with that statement. While you can’t draw many conclusions from it, you can still see patterns that are relevant to predict how the teams will start at the beginning of the season. We saw it last year with Ferrari, don’t forget that. In Haas’ case, they seem to be in contention to be the worst team on the grid. Surely Williams will have something to say though, but so far it is not looking so good for the American team.

Density plot of representative laps


We’ve seen all the laps, but what about the representative laps? To start, I will say that defining what “representative” laps are is quite hard. There is no hard threshold to determine which laps are actually representative of the race pace of each driver. Each driver will have a different thre²shold, but finding it would require an important amount of time and energy to be determined. I, unfortunately, didn’t have that.

To get the laps that I considered to be representative, I went with a more methodical approach that I thought would get us the information that we wanted to see. I first transformed the data to the format of \frac{1}{Y^2}, with Y being each lap time. This transformation allowed me to get fairly gaussian data distribution. Getting this distribution allowed me to work with more traditional methods to filter the laps, and in this case I decided to use the median absolute deviation as my main tool. With the transformed data, I calculated the median for all the laps. Then I decided to set a maximum lap time threshold of

Maximum\;lap\;threshold = median\;time + 2 * MAD (median\;absolute\;deviation)

It is important to mention what the three lines on the density curves mean. In this case, they represent the 25%, 50% and 75% quartiles. The 50% quartile is also known as the median, so this line should move more towards the right in drivers that are closer to the bottom of this chart. The 25% quartile line is the limit that represents the fastest 25% of the laps done by each driver. Drivers with a 25% quartile closer to the left side of the plot did relatively more fast laps than drivers with a 25% quartile line that moves towards the right side of the plot. Times done to the right of the 75% quartile line are the slowest 25% of laps done by each individual driver.


The order of the driver now matters since they are ordered from fastest to slowest based on their median lap time. It may look funny to see Esteban Ocon at the top of the chart, but that’s just what the data said. It doesn’t mean that Ocon will be the fastest driver once the season starts, but it does mean that he had a testing weekend where he did many fast laps. If you take a look at the distribution of laps between him and Max Verstappen, you will see that Max did many more fast laps than Esteban, but his median lap time was “dragged” to the right due to his higher amount of slower laps.

We can see that the chart shows tendencies, which is something that we can take from the 2021 pre-season testing weekend. Mercedes might have had a troublesome weekend, but were still pretty fast and will certainly improve during the next couple of weeks. Red Bull had a great weekend, showing speed and reliability, with Verstappen looking the fastest of the two drivers at the moment. Don’t count Sergio Pérez out though. His median time may have been slower than his teammate’s, but you can see that the times below the 25% quartile line are still among the fastest of the grid.

On the slower side of times, we can see teams like Haas, Williams, and surprisingly, Aston Martin. I think we can agree that expected Haas and Williams to be the slowest two teams, but Aston Martin—formerly Racing Point—had a great season last year and is unexpected to see them struggling at this point. Both Lance Stroll and Sebastian Vettel struggled to get fast laps on their stints, with Vettel having a particular bad time as he was among the drivers with less laps represented in my analysis.

Final remarks

I hope you have enjoyed this simple charts that try to summarize how the 2021 pre-season testing went for the teams. We can’t draw crazy conclusions from the laps that I just showed you, but we can, however, draw certain tendencies that the drivers displayed in the three days of testing.

It is looking like Mercedes and Red Bull will be the fastest teams again—which we kind of expected—but we also saw teams like Alpine and McLaren have a great weekend in terms of both speed and reliability. Last season we saw a very tight midfield, and it’s looking like this season may be even tighter than the last one.

I hope that you have enjoyed this article. If you did, please share it with your friends and let me know what you think in the comments below.


Submit a Comment

Your email address will not be published. Required fields are marked *

Send this to a friend