How money predicts success in Formula 1

The cost of even competing in Formula 1 is astronomical. The teams collectively spend over $2 billion annually — either immensely impressive or depressing, depending on your perspective. About half of those funds come from FOM prize money. The rest is made up by investments from manufacturers, sponsors, and pay-drivers.

Spending priorities are very different at different ends of the grid. For teams like Ferrari, Mercedes, and Red Bull, who can pay their own way, it’s a matter of calculating the optimal amount to invest. At some point, there are diminishing returns, and investing another $1 million may no longer return $1 million worth of value, in terms of performance, brand exposure, and prize money. This determines an optimal spending point. In economic terms, this is the point at which marginal revenue equals marginal cost. You can think of this as a naturally-arising budget cap, albeit one that is far beyond the reach of most teams. The consensus of top teams would suggest that this point is around $500 million annual spending, including money from sponsors and prize money.

For teams lower down the grid, budgets are typically limited by inputs from sponsors and pay-drivers, not to mention the prize fund, which currently gives strong preference to top teams. However, important budgetary decisions remain. For example, is it worth taking a pay-driver over a quicker driver, due to the boost in team performance a larger budget could bring? Typical pay-drivers, such as Ericsson, Gutierrez, or Perez, bring around $20-$30 million. Maldonado brought an incredible £29 million (equal to $46 million) to Williams in 2012, as seen in the leaked invoice below.

As you can see in the graph below of estimated 2014 team budgets, an extra $30 million is a big deal for a team in midfield or below, which is why we see so many well-supported drivers in those teams. Top drivers of course also have sponsors, but top teams can afford to choose their drivers on talent. When the team budget is $300 million or more, an extra $30 million is simply not worth sacrificing half a second per lap or more in driver performance. Indeed, top teams may even spend 10% of their total budget on driver salaries to ensure they have the very best drivers available.

Estimated 2014 team budgets (source), along with prize money estimates (source).

Growth over time

In the early days of Grand Prix racing, the sport was affordable to wealthy privateers, who would buy their own cars from manufacturers such as Maserati and employ a handful of crew members. Today, teams are massive entities, sometimes exceeding 600 personnel. When and how did this change occur?

I searched through a variety of sources for team budget estimates in different eras. Some of the earlier estimates are quoted from team owners. More recently, team budgets have been estimated by Formula 1 magazines and insiders (my full list of online sources is at the end of the article). All of these numbers have to be taken with the caveat that they are estimates by people in the know, rather than exact figures. In some cases, it’s difficult to actually define what encompasses a team’s budget, e.g., when they have subsidiary companies. This has sometimes led to heated arguments.

Plotted below are the maximum team budgets I found in various years. All numbers are converted to US dollars and adjusted for inflation.

The earliest data point is Prince Bira’s estimated cost for his racing program across 1935-1939, which was about $300,000 annually in 2014 US dollars. The Cooper team spent a similar amount in 1959. Beginning in 1968, teams were free to run sponsor liveries. Lotus immediately began running their famous Gold Leaf livery and others soon followed suit. By the mid-1970s, all major teams had sponsor liveries, including tobacco sponsors, oil companies, and alcoholic beverages. A full list is available here. However, overall budgets were still very small compared to today, running around $10 million for a major team.

The early growth of Formula 1 can be seen more clearly by plotting the same data with a logarithmic y-axis. This shows that from 1960-2000, the financial growth was close to exponential, increasing by a factor of 1000 over those four decades. Team budgets grew by a factor of about 6 each decade. At first, this growth was driven by increased investments from manufacturers as they developed new technologies to improve their competitiveness, including improved chassis design and fabrication, as well as the development of vehicle aerodynamics. Growth beyond 1980 was driven primarily by the sport’s expanding global reach and increased investments from sponsors, which fundamentally changed the sport’s economics.

In 1981, the FISA-FOCA war was settled with the signing of the first Concorde Agreement, which granted Bernie Ecclestone control of the sport’s television rights. In this position, Ecclestone sought to rapidly grow the sport’s global appeal and audience, as summarized in this excellent article, from which I quote:

Unique among his contemporaries, Ecclestone had a clear and prescient vision of the future of Formula One, and at its core was television.

In order to realize those opportunities, Ecclestone set about changing the face of Formula One. As both a team owner (Brabham) and head of FOCA, Ecclestone began to mold Formula One into a more telegenic image, in preparation for the coming media explosion he alone foresaw. Jeans and t-shirts were no longer allowed on pit road, quickly replaced by team uniforms. Cars had to be prepared to appearance standards as high as those the teams applied to technical preparation. Gradually, at Ecclestone’s insistence, the rag-tag band of itinerant racers following the summer sun was transformed into a spit-and-polish cadre of professionals, in both appearance and demeanor.

Where previously, most broadcasters had carried their home Grand Prix and perhaps a couple of others, there was nothing like the full-season coverage we’ve come to take for granted. In those countries which hosted a Grand Prix, Ecclestone required that the network which produced the coverage show all the races, offered in exchange for bargain basement rights fees. Country by country, Ecclestone marched Formula One across Europe’s television screens.

Even the rules of Formula One were changed to make the sport more attractive to television and television viewers. Timetables were set and rigidly adhered to by all. And gradually, what Ecclestone had expected became reality: more viewers meant more spectators at the races, and more spectators meant more viewers. Ecclestone had found the entertainment business’s equivalent of the perpetual motion machine, and once the number got large enough, the machine began to disgorge money. Large amounts of money.

Country by country, as the at-home audience became sufficiently substantial, Ecclestone renegotiated his contracts with the broadcasters. What had been virtually free programming was priced in accordance with the number of people who couldn’t get enough of Formula One. The tobacco companies had become involved in F1 long before the FISA-FOCA showdown, John Player being one of the first non-automotive companies to underwrite an F1 team, in the mid-1960s. But with the advent of a major television audience, more came into the sport and all paid more dearly for the increased exposure. Formula One had found its soulmate, and Ecclestone had been the matchmaker.

It was a brilliant strategy, and one Ecclestone executed just as brilliantly. The influx of money transformed Formula One into the highly professional presentation we see today. It also transformed the team owners who stayed the course – and Ecclestone, of course – into multi-millionaires.

By the 1990s, tobacco companies provided the majority of team sponsorship and team budgets broke through the $100 million barrier for the first time. Senna and Prost were earning $10 million ($18 million after inflation) and $9 million ($16 million after inflation), respectively, in 1990. 12 years earlier, the highest driver salary was Lauda’s at $1 million ($3.6 million after inflation). By 1994, Senna was earning a reputed $20 million ($32 million after inflation), comparable to the highest driver salaries in Formula 1 today.

Budgets continued to grow into the late 1990s as the sport’s popularity grew. Global viewership increased by 35% from 1996 to 2001. By 2004, top teams had reached budgets of around $400 million (after inflation). The blanket ban on tobacco sponsorship in 2005 caused a temporary hiccup in the sport’s growth, and budgets were simultaneously reigned in by cost-control measures designed to stop the sport from spending itself into oblivion as teams lower down the grid struggled to keep up with major manufacturers. However, new investments from technology companies and the manufacturers themselves pushed team budgets to new heights before the global financial crisis and recession. In 2007, McLaren’s estimated budget was the highest in the sport’s history at $567 million (after inflation).

Recent changes

After 2007, Formula 1 teams were hit hard by the global recession. This caused the loss of several major manufacturers and major sponsors, leading to greatly reduced team budgets. However, top team budgets are currently increasing again, approaching their highest ever levels. For teams at the lower end of the grid, this has created renewed pressure. In the last three years, this culminated in the collapse of Caterham and HRT, and a last-minute rescue of Marussia. Haas will join the sport in 2016, hopefully boosting the grid to 11 teams. The current grid size of 10 teams is nevertheless normal for the post-1995 era, as shown in the graph below. Prior to that, the grid was filled with many poorly-funded, uncompetitive teams.

It’s also worth noting that the current prize money structure is a clear deterrent to having more than 10 teams. Teams within the top 10 are eligible for both Column 1 money (based on performance in two of the last three years) and Column 2 money (based on performance in the last year), with a team in 10th overall receiving about $60 million. Teams outside the top 10 are currently eligible for neither, receiving only about $10 million from Column 3. This is what sank Caterham and rescued Marussia; their reversal of 10th and 11th positions was worth approximately $50 million in prize money. This system obviously needs amending if the sport is to sustain a larger grid.

I was able to find budget estimates for all teams from 1999-2014. These are plotted below, with individual teams shown as red dots and the blue line showing the yearly average. The WCC-winning team is marked by a green dot in each year.

One thing you may immediately notice is how the points are clustered. In most years, there is a group of points near the top (i.e., the big spenders) and another group of points near the bottom (i.e., the smaller teams). From 2001-2007, there were always 7 or more teams spending at least 50% of the biggest spender’s budget. In 2008, this fell to 5 teams due to decreased spending from Renault and Red Bull, and it has not risen above 5 since then due to the loss of manufacturers Toyota, Honda, and BMW. In 2014, only 4 teams spent at least 50% of the biggest spender’s budget: Red Bull, Ferrari, Mercedes, and McLaren. 2015 also saw a record 5 teams fall below 25% of the biggest spender’s budget. These worrying changes are shown in the graph below.

To summarize, recent years have seen a post-recession recovery in top team budgets and an increase in average budgets, but a broadening disparity between the top half of the grid and the bottom half of the grid. The current prize fund structure awards about $200 million to each of the richest teams, about $60-70 million to each of the poorest teams, and only $10 million to any teams outside the top 10. This goes beyond incentivizing performance and helps maintain the power of a small number of teams. The same could be said of the fact that the poorest teams have no representation in the F1 Strategy Group. All of this creates a very unhealthy economic situation for Formula 1.

Another worrying trend is the reduction in global television audiences since 2008, as shown in the graph below. This has occurred partly due to a shift to pay-TV in many countries. The result is a continued stream of revenue from existing fans in the short term, but also decreased total exposure and the possibility of fewer new fans in the long term. The total failure to embrace online streaming and internet promotions of the sport has surely not helped either, at a time when audiences are shifting to that platform. From 2008-2010, Netflix increased their annual revenue from $1.4 billion to $5.5 billion. In the meantime, FOM’s main accomplishment was removing thousands of videos from youtube that might have piqued the interest of new fans, while providing no means for fans to watch old content. The ship seems to be finally turning around, but it’s slow going.

Money and success

In general, teams with bigger budgets are more successful. Below, I plotted the percentage of total possible constructors’ points scored by each team versus their budget, normalized by dividing by the average team budget in each year. This includes all teams from 1999-2014.

In general, the points follow a sensible relationship. Teams with very very low budgets (less than half the mean budget) are unlikely to score more than 10% of the available points. Teams that are between half the mean budget and the mean budget fare slightly better, with some impressive outliers in this range. Teams that spend more than the mean budget have the best chance of scoring a large number of points, although there is great variability in this range. Interestingly, the four points furthest to the right all fall below where one would expect, showing that money alone does not guarantee success.

The trend line is intended to be illustrative — it is a least-squares two-sigmoid fit to the points, excluding the labeled outliers. Notably, most of the outliers are from either 2009 or 2014. These were both years in which there were massive changes to the technical regulations, showing that in these time periods teams can excel (or fail) based on the ingenuity of their response to the regulations (aerodynamics in 2009 and engines in 2014), rather than raw budgets. When rules are static, the big-spenders seem to pull away.

We can also examine how individual teams have performed. Below shows Ferrari, McLaren, and Williams. With the exception of Ferrari’s spending disasters in 1999, 2010, and 2014, and Williams’s overperformance in 2014 (all labeled as outliers above), these three teams follow the overall trend.

On the other hand, here are the Honda and Toyota fiascoes, which both ended with the global financial crisis. In the case of Honda, the spending finally bore fruit the year after they withdrew, in what has to be one of the most embarrassing managerial decisions in Formula 1 history. Toyota was an incredible wasteful enterprise, burning through over $3 billion in eight years, including the biggest team budgets in 2005 and 2008, without achieving a single race victory. Their staggering failure has been attributed to an unwieldy management structure that required top-down approval for even trivial decisions where time was of the essence.

Predicting the future

Predicting how any team will perform in the next year is a notoriously difficult problem. Many are the preseason reviews from respected and knowledgeable pundits that fall laughably far from the mark just months later. To give one example, in 2010, only one of five BBC pundits predicted Vettel would be WDC and only one of five predicted Red Bull would be WCC.

Using a statistical model, we can explore how well a team’s future performance is predicted by data from the previous year. Two types of data that could be informative are: (i) the team’s points-scoring performance the previous year, and (ii) the team’s budget the previous year.

First, let’s see how points in one year predict points in the next.

There is clearly a positive correlation between these variables, but it’s a noisy one, which is why predictions are difficult for anyone trying to make long-term forecasts, be they gamblers, pundits, or drivers. The points are about equally likely to fall above or below the diagonal line of equal performance. This means that a good initial guess (given no other information) is that a team will perform about the same as they did the previous year. Using this method would correctly predict the WCC in 25 out of 56 years from 1959-2014, and gives R2=0.49 (i.e., 49% of the variation in points can be explained just in terms of the previous year’s points).

Alternatively, we can try to use the previous year’s budget to predict future performance.

Here, we see a very similar relationship between budget and points scored as we did using the current year’s budget above. In general, a bigger budget means a greater chance of scoring more points. However, budget alone is not the best predictor. If we guessed that the team with the biggest budget in the previous year would win the next year’s WCC, we would only have guessed right in 6 out of 15 years from 2000-2014. Meanwhile, a sigmoid fit to the data gives R2=0.43 (i.e., 43% of the variation in points can be explained just in terms of the previous year’s budgets).

By using both the previous year’s points and the previous year’s team budgets in a single statistical model, we can do a better job of explaining the results in the following year. In the above graph, I colored the data points by tertiles (one third of the data in each group). This shows two things. First, a team’s points scoring and budget are related, which we know already. The teams in the lower tertile, based on points scoring in the previous year, also tend to have small budgets. The teams in the upper tertile also tend to have large budgets. In other words, these variables are correlated. Second, the three tertiles all appear to describe roughly the same curve. They just cover different parts of the curve, depending which tertile they are in.

This suggests we can describe the relationship between the dependent variable (points) and the two explanatory variables (previous year’s points and previous year’s budget) in terms of a function with both explanatory variables as inputs. Based on the shape of the curve, and the quality of fit to the data, I chose a double sigmoid function (i.e., a sum of two sigmoid functions, with an overall minimum of 0% and an overall maximum of 100%). As the input I used log10(B) + kP, where B is the previous year’s budget (in multiples of the year’s mean budget), k is a (fitted) constant, and P is the previous year’s percentage of possible points.

This model performs better in explaining team performance than using either of the explanatory variables alone. The best fit, shown by the red line, gives R2=0.59 (i.e., 59% of the variation in points can be explained in terms of the previous year’s budgets and points). The model also allows the explanatory variables to be approximately compared on the same scale. Specifically, with the fitted value of = 0.012, the model predicts that a 10% increase in percentage points scored in the previous year is equivalent to a 30% increase in team budget in the previous year. To put some numbers to this, the following two teams are predicted to be equally competitive in the following year:

  • A team that scored 20% of the possible points and had a budget of $260 million.
  • A team that scored 30% of the possible points and had a budget of $200 million.

While this predictive model performs reasonably well overall, 41% of the variation in team points scoring remains unexplained. Other factors that could explain this variation are obviously numerous and include: quality of team personnel, team upheavals and restructuring, large changes in sponsorship, and quality of the engine supplier. The 2014 results provide a particularly striking example.

In 2014, the engine manufacturer was a very strong determinant for performance. Every Mercedes-engined team overperformed relative to model predictions, while every Ferrari-engined and Renault-engined team underperformed relative to model predictions. On average, Mercedes-engined teams finished about where an equivalent team with three times their 2013 budget would have been predicted to finish. On average, Ferrari-engined and Renault-engined teams finished about where an equivalent team with one third their 2013 budget would have been predicted to finish!

This is certainly not the only time that team performances have been heavily dictated by engine performances. Ford engines were practically unbeatable from 1968-1974, as were Honda engines from 1987-1990, or Renault engines from 1992-1997. The last decade, in which engines were all carefully balanced and usually separated by no more than ~30 horsepower, was a historical anomaly for Formula 1. Now we are back in a situation where engines are at the forefront of technological development and once again they are a key performance differentiator. Whether this is a good or bad state of affairs is a matter of opinion. The biggest concern is that the performance differences may become locked in by development freezes under the current token system, which is a disincentive both to poorly performing engine manufacturers remaining in the sport, and to new engine manufacturers entering the sport.

In 2015, Ferrari seem to have already largely closed the gap to Mercedes, but Renault (and Honda) continue to lag behind for now. This is confirmed by the 2015 predicted performances for each team, based on 2014 budgets and points.

2015 model predictions

# Team Predicted % Points
1 Mercedes 59
2 Red Bull 50
3 Ferrari 42
4 Williams 27
5 McLaren 26
6 Lotus 8
7 Force India 6
8 Toro Rosso 5
9 Sauber 5
10 Caterham 4

Closing thoughts

I hope this analysis gives some new insights into the economic present and past of Formula 1, including where the current problems lie and what it really takes for a modern team to be successful in the sport. In terms of sustainability, there are certainly some concerning trends. These include: the widening disparity between the top few teams and all the rest; a broken prize fund structure that restrains less successful teams and outright eliminates teams outside the top 10; and the increasing costs of customer engines (now up to around $30 million per year).

As always, I’m hopeful for the future. The sport is clearly not at full health, either economically or in its leadership, but the action on track is as good as ever, so the show must go on.

References

The full list of sources I found and used for estimating team budgets is below. In some cases, I found different sources to be contradictory or sources that contained what seemed to be grossly exaggerated budgets compared to all other years. In these cases, I used my best judgment.

Others:

Advertisements

19 comments

  1. Amazing article. Thank you!

  2. Great blog! Do you have any helpful hints for aspiring writers?

    I’m planning to start my own site soon but I’m a little lost on everything.
    Would you recommend starting with a free platform like WordPress or go for a paid option? There are so
    many choices out there that I’m completely overwhelmed
    .. Any ideas? Cheers!, http://goo.gl/xF2piV

  3. Great article. It really shows how the manufacturers spent heavily in the 2000s, while the small teams kept at a more reasonable growth level. I wonder how the top teams will continue this now that audience is dwindling.. Also, it shows why Mosley suggested $400m as a ‘sensible’ budget cap, looking at the long term trends.

    1. *$100m. Maybe start with 4 and work down? 😛

  4. Again, really great article. I would imagine that you would find similar results in many professional sports. One fascinating exception was baseball, as outlined in the book Moneyball. Gotta run but would love your thoughts on that compared to F1.

  5. I love every single one of these posts. Top job!

  6. You, sir, are simply the best.

  7. Zigmundo · · Reply

    I don’t get the prize money estimates. Mercedes won basically everything in 2014 and got less then Red Bull and Ferrari?

    1. This is due to the way in which the prize fund is distributed; it is not purely based on performance in the previous year’s championship. Ferrari take a large bonus off the top (~$90m) for being the sport’s longest lived team (other historical teams receive smaller bonuses). There is also a large bonus fund (~$135m) divided between the teams that have won the most races in the past three years. For now, Red Bull are still the top earner in that pool due to their earlier results.

      1. I see, interesting. They don’t talk about that on TV.

  8. Matt Karshis · · Reply

    Once again, great stuff. I love your articles.
    Speaking of money I saw an article on Autosport with the leaked details of 2014 payments out to the teams. This shows McLaren gets some legacy money and of course Ferrari’s big kickback.
    http://www.autosport.com/news/report.php/id/118955

  9. When I try to make my mind up about how successful I think a team will be, I usually try to find out the technical personnel responsible for a car, like the head of aerodynamics, chief designer, technical director and see which championship-position the team got with them involved. This is also the way I like to evaluate if a team made a good choice employing someone or changing a leadership position (like Mark Smith going to Sauber or Loïc Bigois going to Ferrari or Mercedes replacing Loïc Bigois with Mike Eliott).
    This is a good source:
    http://8w.forix.com/6thgear/designers.html
    The rationale behind is that “the fish always stinks from the head down”, as we say in Germany. With increasing number of team members at the base of the operation the general expertise of the workforce will even out between the teams – it’s the leading positions where competence and incompetence will make the biggest difference.

    I always did some amateurish hand picking and guesswork based on results tables and the Forix site. Unfortunately I don’t have a database with stored results (which you have). but it would be interesting to read some analysis that actually springs from a proper mathematical model to answer questions like: Who is the best project lead? What’s the average gain or loss in championship position that a team gets when employing a particular person? What would haven been the projected championship-position based on a given team lead in a given season, counting all known facts but the season in question – and how did it turn out in reality?

    One might also build a relation to a budget-ranking into the model (unfortunately budget-numbers are not that easy to come by, so I use to skip that step and use a rule of thumb). A well lead team with a good budget will usually be a recipe for success (unless they mess up like McLaren/Honda this year), while the best team will not succeed if key members are not up to the task (Toyota comes to mind) and even pretty good lead will not have any success if the budget is not there (Mike Gascoyne and John Iley at Caterham).

    1. Thanks for the thoughtful comment! I have thought about analyses with respect to the influence of key personnel, such as individuals like Newey or Symonds. Unfortunately, I think it’s a problem that isn’t really amenable to study from the outside of the organization. Whereas cars are driven by only a couple of drivers, with transparent differences in their performance, teams involve the combined efforts of hundreds of individuals with no obvious indication of who contributed what. Even if you reduce the problem to top-ranking personnel, you’ve still got a handful of individuals per team, making it very difficult to correctly identify their individual contributions. This is complicated further by the fact that many teams are likely successful due to strong interactions between certain individuals, rather than the sum of their independent efforts. The results of individual contributions may also occur with some time-lags, which all makes for a very nasty, overfitted model.

      1. For sure it’s all more complicated than that, but that’s the same with drivers too. Some drivers are very sensitive to their car and when it doesn’t fit their style, they sometimes look alarmingly amateurish, while in a properly set up car they excel. Kimi Räikkönnen is one recent example. Also the driver-racing-engineer-connection has a huge influence on performance. But just like outstanding drivers have the skills to adapt their style to the car ot influence the team too follow theirs not their team mate’s needs, the outstanding team leaders have the influence and the skills to form their part of the organization, get the right people for the job and also get the communications right – all of which is vital for a working design and manufacturing process that fits the budget and that sets the upper limit of what a driver can achieve.
        Sometimes that takes a bit longer, sometimes it doesn’t but I came to the conclusion that it’s more likely that a team will improve over the coming years if it’s employing high class lead engineers like Newey+Prodromou, than when it’s emploing a high class driver like say Mika Hakkinnen. It’s also frappant to see that for example when Newey+Prodromou left Williams and McLaren the teams almost instantly suffered – and it’s not just the engine that makes today’s Red Bull look rather ordinary, a lot of that will be caused by Newey taking a step back and Prodromou going to McLaren. You can have very fine drivers, like Alonso and Button at McLaren, and they won’t win anything if the car is no good. With Peter Prom at McLaren, now involved in the 2016 car design from the very start I’m pretty sure the car will look much better as soon as Honda has their problems sorted out.
        So, I don’t think that evaluating teams by they key personnel is anything less reliable than rating drivers. But that’s just a feeling.

        I think really the interesting question is whether a statistical analysis of team key personnel and their influence on success is more accurate than rating teams by their drivers, or if there is a way to connect driver ratings, key personnel ratings and budget ratings to a common model and see how accurate it would predict a team’s success in any given season.

  10. […] large budget is a prerequisite for team success, as my earlier analysis of Formula 1 finances showed. However, money alone does not guarantee the best car. From 1999-2015, the team with the […]

  11. […] large budget is a prerequisite for team success, as my earlier analysis of Formula 1 finances showed. However, money alone does not guarantee the best car. From 1999-2015, the team with the […]

  12. Interesting

  13. […] and fuel loads into account. Making predictions in F1 is extremely difficult, as I found out in a previous analysis of F1 economics. Nevertheless, I think there are significant insights to be gained from a rigorous analysis of the […]

  14. […] my preseason predictions from testing, but advanced impressively across the season, despite their relatively limited budget. One hopes that in the battle for 4th with Williams, they have not diverted too many resources away […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: