Zverev and a Night at Flushing Meadows: Two Five-Setters, One Semifinal, One Recovery Gap

"I watched Shelton-Alcaraz until 3:30 a.m. I did not go to bed."

Alexander Zverev said it in a light, almost amused voice, right after closing out his US Open quarter-final against Botic van de Zandschulp. He is the top seed. He is the only player left in the men's draw who has won a Grand Slam. And on the night before his third career US Open semi-final, he sat in front of a screen and watched a five-set match with two tiebreaks, ending 6-7(5), 6-1, 6-3, 1-6, 7-6(10-7) in favour of Ben Shelton over Carlos Alcaraz.

The ATP Tour's official channel called it a match that "will be remembered for years to come." I read it differently. To me it is a data point about the architecture of a Grand Slam: who watches, until what hour, and where the cost lands.

Numbers whisper. Those who listen hear an entire match.

Context: A draw that has thinned out

The US Open closes the North American hard-court swing, the fourth and final Grand Slam of a long season. For Zverev, the position is clear: top seed, No. 2 in the PIF ATP Rankings, and after his Roland Garros title this season, the only major champion remaining in the draw.

It is his third career US Open semi-final.

That fact needs to sit beside another: he has never won in New York. Three deep runs, no trophy. The Roland Garros title dismantled part of the "never converts" narrative that followed him for years, but it did not answer the hard-court question.

Hard court suits a serve-plus-first-strike archetype best. A tall, big-serving aggressive baseliner has every reason to treat it as home ground. Reaching deep rounds repeatedly without converting here was always a pattern about conversion, not about level.

The draw around him has thinned. Alcaraz, a two-time US Open champion, is out. Shelton and Frances Tiafoe have produced an all-American semi-final, framed by local media as a revival of American men's tennis. Zverev's semi-final opponent is Karen Khachanov.

So the picture has three separate panels: a top seed at his career peak, a draw that has removed the biggest names, and a domestic media story being pushed hard on the other side of the bracket.

For an analyst, those three panels do not say the same thing. The first is about ability. The second is about opportunity. The third is about attention.

Core: What the data actually says

Before trusting a number, ask where it came from. The ATP Tour report supplies no performance metric at all: no first-serve percentage, no points won on first serve, no points won on second serve, no break-point conversion, no winner-to-unforced-error ratio. It is a human-interest report, not a technical one.

The consequence is that any statement about Zverev's form must be derived from outcomes and structure, not from efficiency metrics. In that thin dataset, exactly one fact stands out: Zverev needed five sets in each of his first two matches.

For a tall aggressive baseliner built on his serve, that matters more than it appears. The serve is the last weapon to decay with fatigue. Movement and return decay first. When Zverev tires, he keeps his biggest weapon while the territory he must defend widens. Every extra set is a debt written into the legs, not the arm.

In years of sports data work, I have tracked a similar pattern: when a big server enters a long sequence of matches, his efficiency curve does not collapse on serve. It collapses on return points won. He holds, but he stops breaking. Matches drift to tiebreaks, and tiebreaks are where randomness carries the greatest weight on a tennis court.

A season missing detail is like a match missing stoppage time.

Across the net in the Shelton-Alcaraz match, the only number available is the score: 6-7(5), 6-1, 6-3, 1-6, 7-6(10-7). Read it slowly.

Two tiebreaks in a five-setter, the last ending 10-7, signals a contest where both men protected serve at the decisive moments and margins were created inside a handful of points. Sets two and three went heavily to Shelton, 6-1 and 6-3. Set four went heavily to Alcaraz, 6-1. Three lopsided sets and two tiebreaks describe waves, not stable equilibrium.

The structural read matters more. Shelton is a left-hander with a big serve, a big forehand and flat power. Alcaraz is an all-court player with variety, net skills and rhythm changes. On a fast hard court, flat power plus serve dominance can push a more creative baseliner into defence. The result fits that pattern.

For Zverev, the information cuts two ways.

First, the surface here rewards serve-plus-first-strike, his natural game. That is a tailwind.

Second, the man who beat Alcaraz is a left-handed power server. Historically, that profile is awkward for a serve-dependent baseliner, because spin and ball direction force him to return from less comfortable positions and to hit more backhands through the middle of the court.

But Shelton is not Zverev's semi-final opponent. Khachanov is.

Khachanov hits heavy and flat and builds his matches on attritional baseline rallies. That profile demands rally tolerance and return quality, precisely the two things two five-setters may have worn down. He does not apply pressure through variety. He applies it through repetition, and repetition is the most efficient grinding tool in modern tennis.

Misread one variable and you lose a whole year of direction. The variable easiest to miss here is time. Not time on court, but recovery time between appearances, which the fixed two-week structure of a Grand Slam never extends. No round is postponed. No match is moved. The body fits the schedule, not the other way around.

There is also the points structure to consider. At No. 2 in the PIF ATP Rankings with a major title added this season, Zverev is carrying a large, freshly banked points block. That block must be defended at the corresponding event next season. This is a general principle of the ranking system, not a bespoke judgment. The pressure to hold position does not end when this tournament ends; it moves to a different date on the calendar.

The Khachanov semi-final is therefore a short-term problem sitting inside a longer medium-term one.

Contrarian: What the sleepless night actually reveals

Here I have to be careful, because this is the easiest place to fall into a trap.

The story of Zverev staying up until 3:30 a.m. and sleeping until 1 p.m. is told as a charming anecdote, and media tends to convert it into causation: he stayed up, so he will be tired in the semi-final. One short night does not break an elite athlete. Their bodies are built for such nights, and a long lie-in recovers most of the deficit. Writing that Zverev lost because he stayed up would be inference beyond the data.

Correlation is not causation. The late-night viewing did not create fatigue. It exposed a recovery structure that was already not ideal and, more importantly, it exposed something else: he was confident enough to tell that story publicly.

Read the quote again. He said he could afford it because he was sleeping until 1 p.m. anyway. That explanation suggests he does not regard it as a competitive compromise. For a 29-year-old entering the window where movement-dependent attributes decline first, the cost of incomplete recovery is higher than for a 22-year-old. But that gap is small across one night and large across two weeks.

Two things usually kept separate belong side by side: accumulated load and the quality of the recovery process. Add two five-setters, add one short night, add an attritional opponent. That sum does not produce a linear result. It produces a threshold. Below it, everything looks normal. Above it, the serve is still there but the legs are not.

There is a bigger structural point too. Scheduling Shelton-Alcaraz in the night session and letting it run past midnight does not only affect viewers. It creates an ecosystem. Players still in the draw live in that time zone, train to that rhythm, eat to that rhythm and sleep to that rhythm. The US Open is the Grand Slam most defined by night sessions running into the early hours. That is a player-welfare issue, and it extends well beyond one individual.

At the 2026 World Cup they laughed at my xG. This year they ask me what xG is. A few years ago, measuring load and recovery was a medical-room matter, not a performance-analysis one. Now it sits at the centre of every result projection. One late night at Flushing Meadows does not decide a title. But a scheduling system that pushes players into such nights, repeated across fourteen days, has a measurable effect.

One more angle, on home advantage. Shelton and Tiafoe have the home crowd in New York. Zverev does not. But home advantage in tennis is a smaller variable than people assume, because a tennis ball cannot hear a roar. The real advantage of an all-American semi-final lies elsewhere: it absorbs all the media attention, all the cameras, all the expectation, and leaves the player on the other side of the draw alone.

For Zverev, that is an invisible but real advantage. He is the last major champion standing. He is the one expected to win. Every remaining opponent is measured against a "he should beat him" baseline. That burden cuts both ways: it is recognition, and it is a steeper narrative penalty if he loses. A player nobody expects to win loses nothing by losing. A player everyone expects to win loses a great deal.

One more point worth stating, because it is easy to miss when reading results. Alcaraz losing does not prove the all-court era is over. It proves that on a fast hard court, in one specific match, flat power plus serve dominance won the final tiebreak. Generational change in tennis is not linear. It happens match by match, surface by surface, tiebreak by tiebreak.

Assumptions that may be wrong

I keep this section in every analysis, because data does not defend itself.

Assumption one: that two opening five-setters left a genuine physiological mark. I have no GPS data, no heart-rate data, no load metrics from Zverev's team. I have set counts and inferred duration. If he recovered better than I assume, the entire weighting of this piece shifts.

Assumption two: that Alcaraz's defeat reflects an unusually fast surface or a sub-optimal physical state. This is weak inference, and I include it only to be transparent that I considered it. A far simpler explanation exists: Shelton played the decisive points better. No extra hypothesis is needed.

Assumption three: that 29 is a decline boundary for a tall player. This is a general pattern, not a biological law applied to an individual. Some players preserve movement past 30 through scheduling discipline and footwork technique.

Assumption four: that the draw thinned around Zverev because of early upsets. I only know Alcaraz is out. I do not have the full bracket to verify the rest, so I draw no conclusion.

What to watch next round

Over the past fortnight I have followed the Flushing Meadows matches with one habit: recording set counts for every surviving player, not scores. That shows who is paying the highest price for progress, rather than who is playing best.

The data shows Zverev has paid more than a top seed should in the opening rounds. Whether that bill reaches the Khachanov semi-final will be answered by one metric I will look for: return points won in the first two sets. If that number is abnormally low, the legs have remembered two five-setters. If it holds at a normal level, I read it wrong, and I will say so plainly.

As for Shelton, having just beaten a two-time champion, he walks into an all-American semi-final as the home hero. Tennis history suggests left-handed big servers tend to go deep at the US Open in fast-surface seasons. That is a signal I want to test with data, not with feeling.

Finally, the Flushing Meadows night reminded me of something eighteen years in this industry has taught me. What gets recorded survives. What is only felt fades. That night, one player stayed awake until 3:30 a.m. to watch two others fight. Three days later, people will only remember who reached the final.

The part in between, two five-setters, one short night and one semi-final berth, will vanish from every bulletin.

Unless someone sits down and counts.