70th Anniversary GP: HAM vs VER per micro-sector

After Max Verstappen took a surprising victory at the 70th Anniversary GP, I asked myself one question: “How did that happen?” I did a brand-new analysis that hopefully will allow you to answer that question.

Methodology

I started by obtaining data from the “telemetry” provided by the Formula 1 live app. The data contains information about the speed and position of each driver from the entirety of the race. I then decided to divide the track into 100 micro-sectors, each of about 100th of the length of the track.

After getting this data, I made a spatial analysis to classify each of the data points provided by the timing app into the just created micro-sectors. This meant that I would get the information from all 52 laps divided by micro-sector.

Since the idea was to compare the speed of each driver for each micro-sector, I had to make sure that all the laps that I would use would be comparable between drivers. I couldn’t compare for example Hamilton’s 41st lap with Verstappen’s 41st lap since Hamilton went into the pits at that particular time. To get a valid comparison, I removed all the laps where either driver was in the pits or the lap after going into the pits.

Just to be completely sure that the laps were comparable, I ran an anomaly analysis to remove laps that could have perhaps problems with the telemetry. The analysis, however, found no problems, so I could continue with the rest of the process.

I grouped the data by micro-sector and compared the average speed achieved by each driver in that micro-sector. To get the difference in speed between both drivers, I subtracted the speeds. E.g., in the 88th micro-sector, Verstappen had an average speed of 219.53 km/h, while Lewis had one of 211.93 km/h. After subtracting both speeds, we end up with a delta of 7.6 km/h in favour of Max.
Remember that this average speed was calculated by using the telemetry from all the laps of the race (minus laps with pit stops and anomalies as previously explained).

Finally, I created a map of the track with all of the 100 micro-sectors and coloured them according to the calculated difference in speed between both drivers. Areas with blue colour represent sections in which Max was faster than Lewis, while areas with a green-ish colour represent sectors in which Lewis was faster than Max. The darker the colour, the greater the delta in speed between both drivers. Areas with white or white-ish colour represent micro-sectors in which both drivers were evenly matched.

Just as a note, the finish line is represented with a cross and a circle, so that you can see where each lap starts.

The analysis wasn’t exactly easy to do, but I believe that the time invested in it was worth it.

Delta to leader per sector

I’ve added an image in the carousel above that shows the track according to Formula 1. It shows the corners and DRS information and a couple of other things that my chart doesn’t show (yet…).

We can see a very clear trend. Lewis Hamilton was faster than Verstappen on most of the straights but was usually slower in the curvy sections of the track.

Max had his best mini-sector against Lewis just close to the end of the track, in the micro-sector 88. This is the dark-blue sector shown during turn 15. In this area of the track, Max was on average faster than Lewis by 7.6 km/h.

Lewis had his best micro-sector against max in the micro-sector 81. This is the dark-green sections shown before T1 and just around the position of the speed trap. In this area, Lewis was faster than Max by an average speed of 8.31 km/h.

One interesting take from the chart is how the delta changes quickly coming out of the corners. In pretty much every turn of the track Max outpaced Lewis, but that speed advantage went away during the straights. This most likely represents the power advantage that the F1 W11 has over the RB16.

Final remarks

Max Verstappen won the 70th Anniversary GP due to his speed in the corners. The tire problems that Mercedes had over the race clearly contributed to this phenomenon though, so I wouldn’t jump to the conclusion that Red Bull is faster than Mercedes in every corner. The conclusion is that Max was definitely faster than Lewis in the corners in the last race.

Red Bull however still struggles to find speed in the straights. It is impossible to say if this is caused by a power-deficit or by a more aggressive aerodynamic downforce setup.

I worked really hard to get this analysis done. Anything that requires a brand-new code takes a ton of time, and unfortunately, I don’t have a lot of it at my disposal at the moment. Having said that, I got it done, meaning that I should be able to create analyses like this one in the near future without having to spend as much time as I did during this week.

In any case, 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.

7 Comments

  1. Matt

    Brilliant work mate

    Reply
    • admin

      Thanks Matt, I appreciate the nice comment.

      Reply
  2. Peter

    Very cool!

    A few questions. What kind of anomaly detection did you use? Do you take the different kind of tyres into account (if that is possible)?

    What kind of software did you use? Is the data publicly available (probably not but I’ll ask anyway)?

    Reply
    • admin

      Thanks for the nice comment Peter.
      I’ll try to answer your questions in order.
      1) For anomaly detection I just use an IQR or GESD method to make sure that there are any weird laps on the data. Basically it helps me to remove laps when the safety car was out, or laps that have wrong information because the data sent by the app was wrong.
      2) For the race I don’t believe that it matters which tire they were using. In the case of two drivers who had the same strategy, if one of them picked a better tire than the other and was faster then that’s just how it is. If the question is if I am making any adjustments on the data depending on the tire, then no, it’s just the original data.
      3) I work with R
      4) Yes and no. The data is taken from the F1 live timing app for Android. While the data is technically available for anyone who has access to this app, you need to extrac the information by using more advanced methods (using programming languages such as python and R).

      Reply
  3. Peter

    Thanks for the answers.

    2) Yes, I was thinking later, tyre differences are often expressed in 0.7 secs per lap faster, etc. Your range delta speed is 16 km/h, so it wouldn’t affect it much, if anything. The general trend would still be visible.

    3) I’m a Python person myself 🙂

    Reply
    • admin

      Cool. Python is nice, but I am much more proficient in R. It is a bit more intuitive for data analysis in my opinion, but it’s all a matter of personal opinion. Regardin the tires, in my mind it’s not the best idea to just try to make adjustments based on Pirelli’s data. There is just so much variation that the results would be pretty inaccurate. Some cars actually end up being faster on technically slower tires (mediums vs softs) due to a better temperature management for example. Without that data that the teams actually have, I try to err on the side of caution and do less and show what I believe is fairly accurate information instead of trying to get crazy models with very poor accuracy.

      Reply
  4. Jenner

    Interesting analysis, thank you.

    A key feature I am noticing is Hamilton is faster in and shower out of each corner relative to Max.

    This potentially ties up with what we often hear and read in the media about how the Merc engine is superior where the Redbull chassis under Newey often leads the downforce race.

    Great visualisation.

    Reply

Submit a Comment

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

Send this to a friend