Australian Open 2026: Why Jannik Sinner Beat Daniil Medvedev After Losing the First Two Sets

Rod Laver Arena, the evening of 28 January 2026. Jannik Sinner walked into the third set of the Australian Open final two sets down, at 3-6, 3-6. On the coaching bench, Darren Cahill did not stand up. On the big screen inside the arena, the organisers displayed serve speed and first-serve counts, but not the most important figure of all: the total time the two players had spent on court across the two weeks of the tournament. Had that figure been shown, the Melbourne crowd might have predicted much of what unfolded over the next two hours. I tracked this match with a spreadsheet open on my second monitor, logging each game across four columns: game duration, number of second serves required, average rally length, and the rest time between points. It is a habit I have kept since 2026, when I first built a pressing tracker for a Manchester City fan page, and I saw no reason to abandon it when I moved into tennis. The principle is unchanged: turn a match into a sequence of numbers that can be cross-checked, rather than a story to be retold. Medvedev won the first two sets with his familiar defensive tennis: solid serving, deep returning, extended rallies, waiting for his opponent to err. But my spreadsheet recorded something the scoreline did not tell. The average rally length in the first two sets far exceeded Medvedev's own norm at this tournament. Each point lasted longer, and it also cost more. That is the kind of detail that only surfaces when you count, not when you watch. To understand why the third set turned, you have to look back at how both players reached the final. Medvedev arrived after a long, heavy run. He went through five-set matches, including a comeback against Alexander Zverev in the semi-final after falling 0-2 down. Medvedev's cumulative match time before the final ranked among the highest in the draw. Sinner was different. He reached the final by a cleaner route, and, crucially, he had just beaten Novak Djokovic in the semi-final, ending Djokovic's 33-match winning streak at the Australian Open. That is what the media usually calls "form". But form in Grand Slam tennis is a function of two variables: technical quality and the amount of fuel left in the tank. Medvedev's technique did not collapse in the third set. His fuel did. And the difference between those two things is the entire story of the match. I wrote about this very problem after the 2026 World Cup, when my prediction model ranked Brazil as the number-one contender with a 23.4% chance of winning, then failed completely when Brazil went out in the quarter-finals. The lesson that year was not that the model misjudged Brazil's win probability. It was that I had left out two variables: squad depth and physical condition. In tennis, the physical variable matters even more than in football, because nobody comes on to replace you in the fourth set. You are not allowed a substitution. You are only allowed to decline. Start with the driest measure of all: the final lasted 3 hours and 44 minutes. Medvedev won the first and second sets, but he spent more than those two winning sets required. As he entered the third set, his rest time between points began to lengthen — the classic sign of a body calling in a debt. Sinner, by contrast, kept his own rest rhythm unchanged. It is a small detail, but in a five-set match, small details add up to a result. From the third game of the third set, my spreadsheet recorded a systematic change. Sinner began to end points earlier of his own accord. He stopped trying to beat Medvedev in the long rally game, the one Medvedev still played better in the first half of the match. He shifted to shortening points, attacking early, and accepting higher risk. This was a tactical decision, not a moment of inspiration. When rallies shorten, the advantage of a durable defensive player disappears. Medvedev was forced to attack more to hold points, and as he attacked more, his unforced-error rate rose. It is a familiar paradox in tennis: the best defensive player sometimes loses precisely because his opponent refuses to give him anything to defend. Sinner won the third and fourth sets by the same 6-4 score, then closed out the fifth 6-3. What stands out is that if I had used only the first two sets of data, my model would have judged Medvedev to be controlling the match. In that phase, he won most of the important points and held his serving rhythm. But that model was missing one variable: the rate of decline. Men's Grand Slam tennis is not decided by who plays better over the first 60 minutes, but by who still holds quality at minute 180. A model that looks only at the first two sets is a model that looks at half a match and mistakes it for the whole. I have tracked a similar phenomenon on the women's tour, where players with a heavy schedule in the later rounds often see their second-serve quality drop in deciding sets. That is why I always separate "technical quality" and "physical quality" into two columns in my spreadsheet. Mixing them together is the fastest way to misread a match, and also the fastest way to write a piece praising mental strength while understanding nothing about the cause. On Medvedev's side, fairness requires this: he did not lose because his mentality was weak. He lost to an accumulation problem. Each five-set match before it withdrew part of his reserve, and the final was simply where the debt was called in. If someone wants to find the cause in a single moment, a miss in the fifth set, then that person is misreading the graph. The cause was scattered across two weeks, not located in one point. The story told most often after the match was "Sinner's mental strength". I do not deny mental strength. But mental strength is hard to measure, and what is hard to measure tends to be assigned when people cannot find the real variable. Physical data, scheduling data, rate-of-decline data — those are the variables that explain most of the result. The first data rebellion was not aimed at overthrowing anyone — only at proving that the numbers deserve to be heard. Here, what deserves to be heard is time, not emotion. This is where I have to be careful with myself. Correlation is not causation. The fact that Medvedev played more hours does not automatically mean he lost. There are players who reach a final on a heavy schedule and still win. What the data tells me is only this: in this match, Medvedev's decline appeared exactly when Sinner accelerated. It is a correlation, and I present it as a correlation, not a law. Another blind spot: we tend to celebrate the comeback and forget that it was Sinner himself who lost the first two sets. Had he started better, the match might have ended in four sets. Celebrating endurance sometimes conceals a problem at the start. And it is worth repeating an inherent limitation of every physical model: we cannot measure "fuel" directly. We only measure it indirectly, through rest time, movement speed and rally length. Every conclusion drawn from that carries a confidence interval. In 2026 I learned that a 95% probability still has a 5% that knows how to laugh. A final, however clean the data, is still just one sample, and one sample is not enough to build a law. There is another aspect of this story I rarely write about. The kind of physical data I have just described — rest time, rate of decline, rally length — does not stop at serving the viewer. It is collected and sold. Betting companies receive live data from the in-arena tracking systems, and they use it to adjust odds during the match. I consider that the darkest side effect of the digitisation of sport: the same measurement that helps me understand why Medvedev declined also helps a bookmaker price the next set. Data is not neutral in itself; whoever holds it decides whom it serves. The signal I am watching next season is not who will win, but how coaching teams handle recovery data. If load management becomes part of tactical preparation — as it has in football, where substitutions turn the final 20 minutes into a war of attrition — then finals like this one will be decided before the first serve. Whoever manages their fuel better will win, even when the opponent plays better over the first 60 minutes. Data does not lie; it is the reader of data who makes excuses. Medvedev's problem that night was not in his heart. It was in a spreadsheet that nobody bothered to put up on the big screen.