2019 F1 season: Ferrari’s straight-line speed

Much as been said about Ferrari’s straight-line speed during this year. Since it’s still a hot topic, I decided to analyze the data to see if the rumours about Ferrari’s sudden gains in speed were true. Let’s take a look at the numbers.

Methodology

I worked specifically with the maximum speeds recorded at the speed trap during the qualifying sessions. Why only quali sessions and only use the speed trap data? Well, I have two main reasons. (1) The qualifying session has less variability than the race. During the race, DRS, fuel load, and tire condition, influence the maximum speed a car can reach. During the quali session, all teams run with the minimum amount of fuel allowed, all use DRS, and all make their runs with new tires. (2) People talk about the speed trap ALL the time. You see it on Motorsport.com, Autosport.com, Reddit.com., etc. They talk of speed trap data as hard evidence. They back their arguments with these numbers. Because of that, I decided to see if this “hard” evidence is as hard as people believe it is.

I did not work with crazy models. I did use some modelling tools for my analysis, but I won’t be showing it here. Most of the information and charts that you will see are based on pure data visualization and descriptive statistics.

Data transformation played a big role in this analysis. How can you compare the maximum speeds recorded at Monza, with the maximum speeds obtained in Monaco? I decided to convert the speed data to a relative scale. For every race, I ordered the maximum speeds, normalized them to a 0 to 100% scale, and used those numbers to make the comparisons. Sounds weird? Let me give you an example.

In Spain, Ferrari had the fastest top speed during the quali session. Sebastian Vettel was the fastest, with a recorded speed of 325 km/h at the speed trap. Daniel Ricciardo was the second-fastest driver (324 km/h), while Lance Stroll had the third-fastest speed (323.5 km/h). Since Vettel was the fastest, his speed was transformed to 100%. Ricciardo’s speed was transformed as well, by using the following formula : (Ricciardo’s speed / Vettel’s speed) * 100. The result was 99.69%, meaning that Daniel was 0.31% slower than Vettel. The same was done for Lance (99.53%), and all the other drivers.

Most of the charts only show the top three teams—Ferrari, Mercedes and Red Bull—but some do include the rest of the teams. I figured that most people wouldn’t be interested in talking about a Ferrari-Williams or a Mercedes-Toro Rosso comparison. Was I wrong?

Finally, I filtered the data to contain only the recorded speed for the fastest driver of each team. The idea behind this decision was to show the potential maximum speed a car from each team had during each session.

Just as a note. Normally when I talk about the “fastest driver”, I am referring to their speed over the course of a lap. This means that for me, the fastest driver is the one who had the lowest time throughout a lap/session/race. Since this is an analysis about top speeds, not speed over a lap, I will refer to the driver who had the maximum top speed as the fastest driver.

Relative speed delta to fastest team

Let’s start with the analysis. First let’s talk about the chart and its elements.

• The axes show the relative delta to the fastest driver of the session and the races we saw during the season.
• The dots and lines show the relative delta for each team, while the number shows the speed ranking for that weekend. For example, in Canada, Ferrari had the fastest speed of the quali session. Mercedes was around 1.75% slower than the Scuderia, and were ranked as the 7th fastest team, while Red Bull was 2% slower than Ferrari, and were ranked as the 9th fastest team of the session.
• The coloured lines that you see on the left side represent the raw data points. They are meant to show the distribution of all the points that you see in this graph. Ferrari has a thick, red line around 0%, meaning that most of the time they had a delta of 0% to the leader. Mercedes’ coloured line at 4.5% represents their worst delta of the season, done in Germany.
• The colours seen on the x-axis labels represent the downforce requirements for a particular track. Red means tracks that require high downforce, yellow refers to medium downforce tracks, and green to tracks with low downforce demands.

The thing that first comes to mind by seeing this track is “wow, Ferrari was fast”. Ferrari had the fastest recorded speed in 14 out of the 21 races of the season. Mercedes’ best showing in this category was seen in Singapore, where they were 0.68% slower than the fastest driver of the session. Red Bull was closest to the top in Brazil, where they were only 0.12% slower than Ferrari.

Renault had the fastest recorded speed in France and Great Britain. Toro Rosso also has two appearances here, with them being the fastest in Germany and the United States. Racing Point was the fastest team in Australia and Russia, while McLaren tied for the first position in Singapore.

Relative speed delta to Ferrari

Let’s be honest. Most of you want to know if Ferrari was cheating. You want to know if Ferrari cheated to try to beat Mercedes and Red Bull. So let’s focus on Ferrari and the other two top teams.

This new chart is very similar to the last one, but the speeds were normalized in a different way. Now, Ferrari was used as a reference point. Values below 0 mean speeds that were slower than Ferrari’s and speed over 0 are speeds faster than Ferrari’s. Not much to see actually above zero. Mercedes only managed to have a faster top speed than Ferrari in the first race of the season. Red Bull was slightly faster than Ferrari also in Australia and managed to tie the Italian team in Germany.

The dashed horizontal lines represent the mean delta vs Ferrari over the course of the 2019 season. On average, Mercedes was 1.38% slower than Ferrari, while Red Bull lagged behind even further, being 1.62% slower than the Scuderia.

Relative speed delta to Ferrari (by track’s downforce demands)

As expected, Ferrari demonstrated their superiority, top speed-wise that is, in tracks with low downforce demands. Neither Mercedes nor Red Bull managed to get within a 1% delta in these low downforce tracks. Red Bull was in fact almost 2% slower on average than the Scuderia. On these tracks, the Austrian team never managed to classify in higher than 4th place, showing their inability to challenge for pole positions in low-downforce circuits.

In medium-downforce tracks, the delta is reduced to less than 1.25% for both teams, with Red Bull matching Ferrari in Germany, and staying within 0.1% of the fastest speed in the USA.

In high-downforce tracks, Mercedes’ delta remained pretty much the same as in medium-downforce tracks. The Silver Arrows were able to record a higher speed than Ferrari in Austalia, but generally had a slower top speed than the Scuderia. Red Bull also beat Ferrari in the top speed department in Austria and was 0.12% slower than them at the Interlagos Circuit.

Relative speed delta to Ferrari (before and after the summer break)

That is all good and dandy, but it doesn’t answer one of the most asked questions. Did Ferrari get faster after the summer break? Our chart doesn’t appear to show that. Ferrari got faster compared to Mercedes, but slower against Red Bull.

In Italy, the track where Lewis Hamilton was unable to overtake Charles Leclerc, the numbers show something interesting. While Ferrari was the fastest, Mercedes was ranked as the slowest team of the quali session. Before the Monza Grand Prix, Mercedes had been even slower only in Monaco and Spa. In fact, isn’t Spa the place where the rumour of Ferrari cheating started? It isn’t hard to see why Lewis was unable to overtake Leclerc in the lowest downforce track of the season. Ferrari was racing in a track that fully suited them, while Mercedes had the car with the lowest top speed of the lot. During the quali session, Ferrari had a top-speed of a whooping 349.7 km/h, while Mercedes lagged behind with a top speed of 342.5 km/h. This delta of 7.2 km/h was just too much for Lewis to overcome during the race.

It was strange to see Ferrari taking victory in Singapore, but the Marina Bay circuit is famous for its few number of overtakes. Mercedes had the top speed to challenge for the win, but their struggles with tire wear cost them the opportunity to do it.

United States and Brazil

What about the USA and Brazil? This is where it gets interesting. In the Circuit of the Americas, Ferrari was only the 7th fastest team of the quali session, with a top speed of 330.6 km/h. Red Bull had a top speed of 330.2 km/h, while Mercedes’ maximum speed was of 325.3 km/h.

Why the sudden drop in speed? Ferrari claims that they were trying a higher downforce setup, which sounds ridiculous at first, but let’s think about it. By the time the teams reached the USA GP, the Constructor’s Championship was already sealed. Lewis still wasn’t World Champion, but we all knew that it was a matter of when, not if. After the Scuderia’s resurgence in Spa, Monza, and Singapore, they came back down to earth in Sochi and Suzuka. Ferrari knew by then that their low-downforce philosophy had failed. Why not start thinking about next season? Ferrari may have failed to take pole position in the USA, but they did it by only one-hundredth of a second, meaning that they still had that top speed.

In Brazil, Red Bull almost matched Ferrari once again. Mercedes was the slowest team once again, indicating their preference for high-downforce setups. This is the only place where the delta between Ferrari and Red Bull may seem like an anomaly. I ran the statistical tests to determine this though, and statistically, it is not an anomaly.  Why was Red Bull so close to Ferrari this time though? It’s hard to say. I can only speculate that Red Bull had a car that was efficient enough in high-downforce circuits to decrease the downforce levels, allowing them to increase their top speed. This, combined with them using their 5th internal combustion engine of the season, compared to Ferrari and Mercedes’ 3rd, gave them the edge at the Interlagos Circuit.

Relative speed delta to Ferrari (before and after the summer break)… again?

Wait, the same chart? Not really, but almost. Some of you may argue that the average delta after the summer break is influenced by the USA and Brazil GPs, after FIA’s investigation into the supposedly illegal trick done by the Italian team. Because of that, I decided to redo the chart, but without the last three races of the season.

The numbers change a bit compared to the last chart. Mercedes’ delta to Ferrari increases to 1.62% over the previous 1.53% value. Red Bull’s delta in the meantime goes from 1.71% to an average of 1.96%. This is to be expected, since as we just saw, Red Bull was not far from the Scuderia in COTA and Interlagos. 

I did statistical tests on the data and the results were not surprising. Ferrari was faster on average than both Mercedes and Red Bull, before and after the summer break. Meanwhile, Mercedes and Red Bull were not statistically different from each other. This doesn’t mean much in real life, since at the end of the day, Mercedes was actually faster than Red Bull. With more data points, we may be able to have the expected statistical significance. Regarding Ferrari before and after the summer break, the data shows no statistical difference. This means that at least according to the stats, Ferrari was just as fast before and after the summer break.

Relative speed delta to Ferrari (2nd and 3rd fastest teams)

What if Ferrari didn’t get faster against Mercedes and Red Bull, but they did against the second and third fastest teams in quali? Yeah, it was a far fetched theory, but I still decided to give it a go.

The chart here shows all the races where Ferrari had the fastest speed in quali. The black dashed lines show the average delta for the second and third teams. Once again, I did not find any significant difference between the deltas before and after the summer break. If anything, the second-fastest teams gained on Ferrari during the second half of the season.

The third-fastest team was on average just as fast after the summer break than before it. Again, an expected behaviour.

Relative speed delta to fastest team (with anomalies)

The final chart of the day, just Ferrari’s numbers. As I said before, I did some testing to detect outliers. This is actually one of the results’ that I got after analyzing Ferrari’s data.

The blue circle represents an anomaly found in the data. Outlier detection is not easy though. Depending on the method that you use, the results may vary. In this case, I used four different methods. After taking information from all of them, I generated this chart. The German Grand Prix is the only data point detected as an outlier, since Ferrari was almost 4% slower than the leader. Let’s remember that during this quali session, Sebastian Vettel retired in Q1, while Leclerc was unable to set a final run in Q3 due to mechanical issues. In a few words, we were unable to see Ferrari’s true top speed in Mercedes’ home race.

Finally, it’s important to notice that neither the Russian GP, nor the USA GP, are considered outliers.

Final remarks

Was Ferrari cheating? I can’t say for sure. The only way to know with 100% certainty is by actually taking a look at Ferrari’s mechanical components.

Statistically speaking, I do not see evidence to believe that the Scuderia was doing something illegal. The results weren’t good for the team in the first half of the season, so the victories in Spa, Monza and Singapore certainly raised suspicions. However, they didn’t suddenly gain speed after the summer break as some believe. The reality is that Ferrari always had the car with the fastest straight-line speed. 

I want to state that like any study, this analysis has limitations. I only focused on the straight-line speed in quali sessions and only took the data from the speed trap. A study that uses more data may, or may not, get different results than the ones shown here.

Just to finish,  and before anyone says it, no, I am not trying to defend Ferrari. I always do my best to be unbiased when doing these analyses. This becomes easier when the analysis doesn’t include my favourite team, like this time. Which team is my favourite you may ask? I won’t say, but it’s neither one of the three teams that were analyzed here.

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.