Tennis on clay
Some statistical insights in view of the Olympic tournament
With Alcaraz winning Roland Garros, the clay season should have been over; Carlos is attacking the first position in the ranking, with Sinner trying to defend it. But this year, an Olympic year, clay season is not over yet: the Olympic tournament has just begun in Paris.
But what can we expect from another big tournament on clay? How should the player behave, in order to be most effective on this surface, unusually getting back to it after grass, after Wimbledon? Well, as we know clay makes the ball slower, with a higher bounce. This makes rallies generally longer and benefits players who use more spin, as we have seen in the Sinner-Alcaraz semi-final match of June, 7th in Paris, with Alcaraz prevailing in the long run. Conversely, big servers may have issues, because their best shot, especially if purely based on speed, produces fewer points.
Let’s think about Andy Roddick: he got to at least semi-finals in every Slam tournament except Paris, where his best result was round of 16, and he got there only once, in 2009. Or, at an even higher level, we can think about one of the greatest American players, Pete Sampras, who won 14 Slams, and never went beyond semi-finals in Paris, again, getting there only once, in 1996.
But it’s not only the speed of the ball or the height of its bounce which makes a match on a clay court different: being successful there, as Rafa Nadal has shown better than anyone else, requires patience and a strategic mind, minimizing risks, and avoiding unforced errors at all costs. In fact, considering that getting to a winner shot is harder, it becomes more important to avoid distractions and unforced errors.
After these general considerations, we wonder what data and statistics can tell us: let’s have a look at the dataset that Jeff Sackmann put together, trying to infer which patterns of play are more likely to lead to a win on clay and how these patterns evolved over time. Due to better data quality, the starting point of our analysis will be 1990, so we will consider the 1990–2024 period (which comprises more than 30K matches), splitting it into three big phases, that we will compare to each other.
Early nineties: before Kuerten
Our first period will last from 1990, which sets the start of our analysis, until the year before the first win by “Guga” Kuerten in Paris (1997): we will focus then on the 1990–1996 period. At that time, clay was the reign of very regular players, like Sergi Bruguera (winner of Roland Garros in 1993 and 1994), Jim Courier (who won in 1991 and 1992), or Thomas Muster, who gained the title on the Philippe Chartier court in 1995 against Chang in the final.
We are tempted to expect that matches could be won with fewer aces and a smaller percentage of points won on first serve, with respect to what would happen in the following years and decades. But is it really the case?
It seems to be, for what concerns the absolute number of aces. In fact, in the 1990–1996 period players who won a match on clay had an average of 3.5 aces, while the same statistic grows to 4.6 (+31%) in the 1997–2024 period. Similarly, if we compare the average number of aces of the player who won with the average number of aces of the player who lost, we have a difference of 0.83 in 1990–1996, and of 1.15 (+38.6% relative increase) in 1997–2024.
Conversely, there is a smaller evolution in terms of what happens when the first serve is in: the players who won a match on clay in 1990–1996 had, on average, a percentage of points won of first serve which was 10.8% higher than their opponent. The same statistic is higher, but only marginally (11.2%, with a relative increase of 3.7%) in the period from 1997 until today.
It seems than that also big players in the early nineties knew how to take advantage of their first serve, but they did it in a less immediate way. As Agassi once said, he was very happy about his (not so quick) serve, because, being very accurate, it frequently allowed him to play an easy first shot. And that, said the kid from Las Vegas, was often enough to win the point.
In 1997, something happened: for the first time, Gustavo Kuerten won Roland Garros. And his style of play was surely different, with respect to Bruguera, Courier and Muster.
From Kuerten to Nadal
From 1997 until 2004, it is quite reasonable to say that Guga Kuerten is among the most representative winning players on clay. His playing style is more aggressive, both for what concerns the serve and the willingness to approach the net. At the same time, also players like Guillermo Coria or Juan Carlos Ferrero won many matches on clay, and they placed a higher emphasis on reducing unforced error and playing a wise match from a tactical point of view.
Also, in this period we have the first years of Roger Federer: he was not yet dominating the circuit and fighting epic matches with Nadal, but he was growing fast, and he was attacking from the baseline, approaching the net if needed. We may then assume that the 1997–2004 period, before the first Roland Garros win by Nadal, has been characterized by high variability in the characteristics of winning players. Is this really the case?
Let us consider the number of break points in a match, which can help giving us a sense of the kind of matches that were played (very tidy or more dynamic).Let us then compare the ratio between standard deviation of the number of break points (which measures how much the values of a distribution deviate from their average value) and average number of break points: we have a value 0.48 for the 1997–2004 period; slightly higher (+4,3%, in relative terms) than the one we observe in the rest of the dataset, that is before 1997 and from 2005 on (0.46).
Also, if we compute the same ratio (that is standard deviation with respect to average) for the number of break points saved by the winning player, we obtain 0.85 in 1997–2004, 2.4% more than in the other periods (0.83).
Some more uncertainty then, and higher variability from 1997 to 2004. And what after 2004? Well, after 2004 the strongest player of all time on clay, who would win Roland Garros fourteen times, went on stage: Rafa Nadal.
The Nadal era
From 2005 on, Rafa Nadal has been the player to beat, if you wanted to win a tournament on clay and, as Roger Federer knows very well, it was pretty hard to do it (only in 2009 Federer managed to win Roland Garros, completing his career Grand Slam, and in 2009 Nadal was defeated by Robin Soderling, so he did not need to face him).
We may say that the impact of Rafa on tennis, especially on tennis played on clay, was so relevant that the dynamic of big matches seemed to progressively change after that. An increasing number of players seems to adopt the tactics of heavily using top spin, reducing risks and accepting a long and tactical fight.
Again: let us verify if this intuitive impression is confirmed by data. Indeed, the average duration of a match on clay was around 100 minutes, both in the 1990–1996 period (100.3 minutes) and in the 1997–2004 period (99.0 minutes). From 2005 on, it increased significantly, reaching an average value of 109.7 minutes, with a relative increase of nearly 10%.
Longer matches then…and more tactical? It would seem so, even if the increase is not equally strong. The number that we could consider here is the percentage of points won on second serve by the winning players, that is the percentage of cases in which, even without a strong serve in, the winning player can overcome his opponent. This statistic has increased over time: 54.8% on average in 1990–1996, 55.3% in 1997–2004 and 56.3% from 2005 on.
Thinking about the latest edition of Roland Garros we would be tempted to say that, very recently, net play is becoming more popular also on clay, possibly because of the very creative way to approach the match by Carlos Alcaraz (and also Sinner is a good net player, even if he usually prefers to rely on his unstoppable rhythm from the baseline).
Yet, this trend seems to emerge just now and it would be hard to sustain that we already have enough data to reliably try to evaluate it. Yet, of course, the evolution of playing style is in fact continuous and, maybe, in some time, we will have the chance to analyze statistics again, looking for confirmation or denial and, essentially, quantification about the emergence of this and other new patterns.
