The Second-Serve Rhythm: What Australian Open Data Reveals About a Tennis Era Changing Its Guard

Rod Laver Arena, a January evening. A fifth set stretching into its fourth hour. The score 40-30 on Daniil Medvedev's serve. He tosses up a second serve — 168 km/h, slice, aimed at the T. The electronic line-calling system flashes a message: the ball is out by 1.2 millimetres. The stands fall silent for two seconds.

In those two seconds, I realised the number I had been tracking for three weeks of the tournament — second-serve points won — had just shifted before my eyes. It changed not because the player hit better or worse, but because of a line. Numbers whisper. Those who listen will hear an entire match.

I sat in the stands with a stack of pre-match stat sheets, and in my hands was the serve-rhythm model I had built over years. That model says that in a fifth set, at this level, the second-serve points won by a top-eight seed typically hovers between 52 and 55 percent. When it drops below 48, the set has effectively changed hands. And at that moment, on court, the number sat at 47.

What I want to tell in this piece is not a win or a loss. I want to tell of a larger shift, one that Australian Open data in recent seasons is laying bare: the serve — the weapon long seen as the absolute insurance of modern tennis — is losing its exclusive power, and people only notice when they read closely the numbers that seem so dry.

Context: When method determines what you see

Before you trust a number, ask where it was born. That is the first principle I apply whenever I open a tennis stat sheet. At the Australian Open, at least three different data sources coexist, and they do not always tell the same story.

The first is the electronic line-calling system, which the Melbourne tournament has used for years to determine whether a ball is in or out. The second is the official ATP and WTA statistics team, who sit up high and tap in every point, recording each serve, each net approach, each double fault. The third is independent sports data companies, collecting tracking data and modelling positions.

These three sources measure the same match but with three different definitions. A serve counted as "in" by the electronic system may be logged as a "fault" by the stats team if the line calibration differs by millimetres. A player's first-serve points won can differ by 2 to 3 percent depending on whether a rally is marked a "winner" or an "opponent error".

That is why I always note the data version in every analysis. A season missing detail is like a match missing stoppage time: you feel you have watched it all, but in truth you have missed the decisive part.

From my experience tracking matches in Melbourne over many years, I have noticed something interesting about geography. The hard courts at the Australian Open are faster than Roland Garros but slower than Wimbledon in some recent seasons, and court speed is a variable that directly affects the value of the serve. When the court is fast, the serve wins more. When the court slows, the serve loses its edge, and the match returns to long rallies.

Home court is not just geography, until it disappears. In Melbourne, home advantage largely belongs to Australian players, who are fuelled by the crowd. But in the data, that advantage is far smaller than people imagine. I once measured the service-points-won rate of home players against visiting players in first-round matches, and the gap was only 1 to 2 percent — enough to create a few moments, not enough to rewrite a career.

Core: A chain of evidence on the decline of serve power

Over roughly the past fifteen years, the scoring structure of elite men's tennis has changed in a fairly clear direction. If you take the first-serve points won of the top ten players and chart it by season, you see a nearly flat line, oscillating around 75 to 78 percent. But if you take second-serve points won, you see a steeper line, and more importantly, one that diverges more sharply between players.

What does this mean? It means the first serve has become a universal skill. Almost every player in the top 100 can hit a first serve at 190 km/h or more with acceptable accuracy. The gap between the best server and the average server on the first serve has narrowed. As a result, the competitive edge has shifted to the second serve — where technique, psychology and the ability to withstand pressure create far greater differences.

Take a concrete example. In a recent season, a top-seed player reached a first-serve points won rate of 79 percent — among the best in the tournament. But his second-serve points won rate sat at just 51 percent. That means every time he missed a first serve, he stepped into what was essentially a coin toss. And over a long set, the number of times he faced a second serve was high enough that the 51 percent became an exploitable hole.

Conversely, there are players whose gap between first and second serve is very small. Such a player might reach only 74 percent on the first serve, but 58 percent on the second. Overall, he wins more service points than the one with a big first serve but a weak second. This is the paradox the data lays bare: the big first server is not necessarily the best hold-of-serve player.

In many matches in Melbourne, I recorded a repeating pattern. Player A serves first at an average of 205 km/h, winning 80 percent of first-serve points. Player B serves first at 195 km/h, winning 76 percent. On the surface, A looks superior. But when the match enters a tie-break, the pressure pulls both players' first-serve-in rate down. At that point, the second serve becomes the only shield. And if B has a better second serve, B is usually the one who wins the tie-break — despite A having the stronger first-serve weapon.

This is the point many viewers overlook. They remember the aces, the first serves that tear the court open. But they do not remember the second serves put into play under pressure, the balls the player had to strain to avoid losing the point. The data remembers. The data always remembers.

Another notable variable is the second-serve points won rate when facing break point. This figure is separate from the ordinary second-serve points won rate, and it measures something I call "the ability to withstand pressure on serve". In one Australian Open season, I tracked the top eight seeds and found this rate ranged from 44 to 62 percent — an enormous gap. The player at 44 percent was almost certain to be broken at decisive moments. The player at 62 percent possessed a mental weapon no stadium scoreboard prints.

I once witnessed a quarter-final lasting four and a half hours, in which a player saved eleven of thirteen break points. His second-serve points won rate in that match was 57 percent, well above his own tournament average. That was one of those cases where I had to be careful: was this genuine progress, or just a lucky night? The answer lay in cross-checking that figure against three other matches by the same player in the same tournament. When I did, I found he sustained above 54 percent across four consecutive matches. That was no longer luck. That was a forged skill.

Another trend the data shows is the rise of long rallies in deciding sets. In semi-finals and finals, the number of rallies exceeding nine shots rises significantly compared with the first round. This reflects a tactical reality: when both players serve well, points are no longer decided on the serve, but on the third shot, the fourth and beyond. And when points extend, movement, positioning and rhythm become the key factors.

In that context, one begins to see the value of players who do not have the strongest serve but possess excellent fitness and return ability. They turn the opponent's serve from a weapon into an opportunity. They stand deep behind the baseline, accept the positional disadvantage in exchange for time, and use that time to return the ball with high spin, forcing the opponent to hit one more shot.

That is why I was not surprised to see some top players sustain a return-points-won rate above 30 percent throughout a tournament. The figure of 30 percent sounds modest, but in tennis it means that out of every ten return points, the player wins three. In a set, that is enough to create a break point, and one break point is enough to create a set.

Misreading one variable is like losing direction for an entire year. I once underestimated the importance of the first-serve-in rate in deciding sets, and that skewed my prediction model for a whole season. Since then, I have separated the data by set, by pressure phase, and by court type.

One further interesting thing is the change in how players use the second serve. In the past, the second serve was usually a safety shot — players reduced speed, increased spin, pushed the ball into the middle of the court to avoid double faults. Today, many top players accept higher risk on the second serve. They use slice or topspin to create a hard-to-read shot, even attacking directly. This is a change of philosophy: the second serve is no longer a place to defend, but a place to attack.

But this philosophy has a flip side. When you attack on the second serve, you also raise the risk of a double fault. And a double fault in a deciding set is one of the fastest ways to lose a match. I have seen players lose titles because of a single double fault at the wrong moment. That is the price of boldness, and the data shows not everyone can pay it.

Contrarian angle: Correlation is not causation

Here I must address what every data analyst must repeat, even when it makes the piece less appealing. Correlation is not causation.

When you see a player with a high second-serve points won rate who also wins the tournament, you are tempted to conclude that a good second serve leads to a title. But the relationship is far more complex. Perhaps winning — that is, winning many matches, playing many sets, facing many opponents — created the conditions for the player to accumulate a high second-serve rate, not the other way around. Perhaps a third variable — such as mental stability, or the quality of the backhand — is the common cause of both.

This is a trap I once fell into. Years ago, I built a prediction model based on the second-serve rate and was confident it would accurately forecast the champion. The model failed in two consecutive tournaments. Not because the data was wrong, but because I confused a descriptive metric with a predictive one. A descriptive metric tells you what happened. A predictive metric tells you what might happen. These two are different, and confusing them is a common mistake.

There is another aspect I want to put on the table: the role of technology in shaping data. The electronic line-calling system has changed how matches are officiated, and it has also changed how data is generated. When a ball is called out by 1.2 millimetres, that is not just a referee decision — it is a data point. And when thousands of such decisions accumulate, they shape what we call a "trend".

Here I have a view that may be controversial. Using technology to determine in or out at the millimetre level has changed the nature of tennis, and not for the better. I do not deny the accuracy of the technology. I question its consequences. When a player knows every ball will be measured to the millimetre, he tends to hit more safely, avoid the lines, and choose lower-risk options. The attacking instinct — the thing that makes tennis beautiful — is narrowed.

This is a stance I share with my view on officiating in sport generally: technology is turning the operator into an editor. In tennis, the electronic line-calling system does not merely correct the umpire's error — it rewrites the player's tactics. And that, to me, is a loss that is hard for data to measure.

Of course, I must acknowledge the limits of this argument. I do not have data to prove that electronic line calling reduces the number of down-the-line attacking shots. That is a hypothesis, not a conclusion. And in my work, I try to distinguish clearly between the two.

Another counter-intuitive point concerns the mental factor. People often say a player wins because of "mental steel". But mental steel is not a directly measurable variable. When I look at data, I do not see "mentality". I see the second-serve points won rate under pressure, the tie-break win rate, the break-point save rate, and the double-fault rate at decisive moments. These numbers are substitutes for what people call mentality. And they are far more objective.

This does not mean psychology does not matter. It means psychology, if it truly exists and has an effect, must leave traces in the data. If it leaves no trace, then perhaps we are assigning it a power it does not have.

Assumptions that might be wrong

Before the final section, I want to spend a paragraph on the assumptions in this analysis that might be wrong, because I have learned that acknowledging the limits of data is a way of respecting the reader.

First assumption: I take the official tournament statistics to be reliable. But as noted above, there are multiple data sources, and they can differ. If the source I use has a systematic error, then part of my conclusion is wrong accordingly.

Second assumption: I take second-serve points won to be a meaningful metric across all court surfaces. But this may not be true. On grass, the second serve has a different value than on clay. Pooling data from different surfaces may obscure important differences.

Third assumption: I take my sample size to be large enough to draw conclusions. But in tennis, the number of top-level matches each season is finite. A few unusual matches can skew the whole picture.

Acknowledging these assumptions does not weaken the analysis. It makes it more honest. And for careful readers — those who want to know where a number comes from before trusting it — that honesty is a sign of respect.

Takeaway: A signal for the next round

When I left Melbourne, carrying in my head the numbers of three weeks of the tournament, I asked myself what would change in the next season. The current data gives me a few signals.

First signal: serve power will continue to disperse. As more players understand that a first serve alone no longer insures a set, they will invest more in the second serve. We may see more players using the second serve as an active attacking weapon, rather than a defensive option.

Second signal: return ability will become a more highly valued skill. Players who can stand deep, read spin, and return the ball with high difficulty will have a greater edge over those who rely only on serve power.

Third signal: the role of technology in shaping tactics will remain contentious. As electronic line calling becomes more accurate, pressure on players to avoid the lines will grow. And that may lead to a safer brand of tennis, but also a less bold one.

I am not sure I am right about these signals. They are progressive judgements, not absolute conclusions. But if I had to bet on one thing for next season, I would bet that the next Grand Slam champion will be the player whose second-serve points won rate sits in the highest group — not necessarily the one with the strongest first serve.

That is what the numbers are whispering to us. And in this sport, those who listen to the numbers are often those who understand the match more deeply than those who only look at the scoreboard.

Transfer value is a story, but data is the signature. In tennis, it is the same. Aces shine under the lights, but the second serves under pressure are what write the champion's name.