The Empty Data Sheet at a Major: The Trust Trap in Tennis Analytics

On the second monitor, every cell returned the same value: N/A. That night in Brisbane I sat in front of two screens — one holding the match log of a major, the other holding the tracking sheet I build for every match day. The first-serve-in column was empty. The points-won-on-first-serve column was empty. The break-point conversion column was empty. The winner-to-unforced-error column was empty. The match still happened. The winner still advanced. Only the data layer above me returned an empty set, and everything below it was forced to inherit that emptiness.

That moment was not dramatic. It was quiet. But it forced me to face a question the sports-analytics trade usually avoids: what happens to the conclusion when the evidence disappears?

I run my work in two layers. The first layer is extraction — recording the events, people, timing, and raw information points of a match or a report. The second layer is deep analysis — turning those information points into verifiable judgments. My iron rule: every conclusion at the second layer must trace back to at least one information point at the first layer. When the first layer is empty, the second layer has nothing to hold onto. It has only two choices — return the gap, or invent a plausible story. And most of the sports-media industry takes the second choice without ever naming it.

That is why I am writing these lines. Not to recount a night of lost data, but to talk about something more dangerous: an analytical system so confident it can no longer tell evidence apart from a story built to fill the void.

When the extraction layer collapses

In tennis, the extraction layer looks far simpler than in other sports. A match has a server, a returner, a score by game and by set. But that simplicity is an illusion. To judge a player I need to know what percentage of points he wins on first serve, on second serve, how many break points he saves, how many chances he creates and converts. Those numbers do not appear on their own. They come from a chain of devices and people: the electronic line-calling system, the statisticians, the aggregation software, and only then the analyst.

Hawk-Eye was first introduced at a Grand Slam at the 2026 US Open, and electronic line calling has since become default infrastructure at the majors. When that infrastructure runs correctly, I get a data row for every point. When a sensor grid fails, when a logging file corrupts, when a provider changes format without warning — the extraction layer returns a gap. And that gap flows down into my analysis layer.

What is worth noting is that the gap does not announce itself as a gap. It wears the appearance of neutrality. A spreadsheet full of N/A still looks tidy, still looks professional, still looks ready for someone to read and assign meaning to. The risk is not that the data is missing. The risk is that the reader is never told it is missing.

I have been on the other side of this mistake. In 2026, when I was sixteen, I wrote analysis posts for a Manchester City fan site. In the December match against Bournemouth, I pulled pressing data from a public source and found the opponent touched the ball just three times inside the box across ninety minutes. I wrote a two-thousand-word piece using expected goals to prove that Pep Guardiola's side was not winning on luck. It was shared, reaching fifteen thousand reads in a day. But what I remember most is not the read count. It is the feeling of certainty. I believed I was right because I had a spreadsheet. I never asked whether the spreadsheet was complete.

Data does not lie; it is the reader of data who makes excuses.

The following year I paid for that certainty. Ahead of the 2026 World Cup I built a prediction model from the historical data of six major tournaments, using Elo ratings and qualifying records. The model ranked Brazil as the number-one contender with a 23.4 percent chance of winning. I wrote a piece declaring that the data had revealed the champion. Brazil were knocked out by Belgium in the quarter-finals. France, whom my model ranked only fourth at 11.2 percent, lifted the trophy. In 2026 I learned that a 95 percent probability still has a 5 percent that laughs.

The lesson was not that the model was wrong. The lesson was that I presented a model missing variables as if it were complete. I lacked data on squad depth and the mental state of stars, but I did not write down what was missing. In the month after the tournament I gathered each player's club minutes before the tournament, added them to the model, and rewrote the entire algorithm. Since then, every analysis I write ends with a dedicated section: what the model does not see.

The chain of evidence and the trap of completeness

Back to tennis. When I have complete data, the reading becomes far clearer. A player can win a match with a first-serve-in rate of only 55 percent if he saves seven of eight break points and keeps his unforced-error rate low. Another player can lose despite winning more total points, because he wins points steadily in unimportant games and collapses in the decisive ones. Aggregate metrics cannot tell that story. Contextual metrics can.

I have been tracking matches this way for nine major seasons. Based on my experience following matches, there is a notably recurring pattern at hard-court events: players who win through serving tend to have a narrower range of results, while players who win through returning have a wider range and depend heavily on whether they can break the opponent's serving rhythm in the first two games of each set.

That pattern is not a law. It is a hypothesis I re-test every week. And here is the crux: a hypothesis is only valuable when I have data to refute it. If my sheet is empty, I cannot refute myself. An analyst who cannot refute himself is an analyst on the road to becoming a propagandist.

There was a moment in 2026 when I learned the value of a clean sample. When the Premier League restarted after the pandemic in empty stadiums, I compared one hundred pre-pandemic matches with fifty post-restart matches. The results forced me to rewrite many assumptions: passes before a contested ball fell from 9.8 to 11.6, meaning teams played slower and more cautiously without crowd pressure. Expected goals from set pieces fell 14 percent, while free-kick conversion rose 18 percent as the psychological factor was removed. The season without spectators was the cleanest laboratory football has ever had. And from the empty stadiums, I could hear the breathing of the match.

What I carried from that study into tennis is simple: when the environment changes, old data loses value in measurable ways. A model built on data from before a rule change, a surface change, or a playing-condition change needs recalibration before reuse. Skipping that recalibration is another form of empty data — not empty because it is missing, but empty because it is outdated.

Another example I still retell as a milestone. At Euro 2026, after Denmark lost 0-1 to Finland in the opener following Christian Eriksen's incident, veteran reporters in the newsroom where I freelanced wrote pieces criticizing coach Kasper Hjulmand for a lack of tactical courage. I pulled the data and saw Denmark generated the highest total expected goals in the group stage — 3.6 — behind only France and Spain. I wrote a rebuttal, using pressing and shot-creating-action numbers to argue that Denmark's performance was not poor, only unlucky. The editor-in-chief, a man of the eye-test school, killed the piece for going against the common feeling. The following week, Denmark reached the semi-finals. The piece ran, and became the most-read article of the month with forty-five thousand views.

The first data rebellion was never meant to overthrow anyone — only to prove the number deserved to be heard.

The correlation trap: when correct data still leads to the wrong conclusion

Here I must warn myself. Having complete data is still not enough. In tennis, false correlations are everywhere, and they are more dangerous because they look like evidence.

Take an example I nearly fell for. A player wins many matches when his first-serve-in rate is above 65 percent. Read quickly, we conclude first-serve-in is the key to his success. But if he faced only weak opponents in the early rounds — when he was relaxed and hit his first serve harder — then the high first-serve rate is a consequence of facing weak opponents, not a cause of victory. Correlation is not causation. In sports data this is not a mere aphorism. It is an operating trap, and it catches even the most careful data readers.

I see the same thing with mental metrics. Players reputed to have nerve at break points tend to have high break-point save rates. But the save rate depends heavily on the quality of the opponent's second serve. When the opponent's second serve is weak, saving break points becomes easier, and the player looks more clutch than he is. What is called nerve is sometimes just a lucky division.

The second blind spot is small samples. A player wins four of his last five decisive matches. That 80 percent sounds impressive until you realize the sample is only five matches and its confidence interval is so wide it is nearly meaningless. Sports media hates confidence intervals because they do not generate headlines. But an analyst who drops confidence intervals is an analyst selling a certainty he does not have.

This is where I deliberately go against the crowd. When the whole market hypes a rising player, my question is not whether he is good. My question is how many matches his record rests on, what tier his opponents were, and how much of it came from repeatable key points. If the answer is that his record comes mostly from converting key points well, I flag it as a repeatable signal. If it comes from opponents collapsing on their own, I flag it as luck. The difference between the two cases is not in the final score. It is in the structure of the points.

I have to admit one thing: there are phenomena I cannot explain with my current data. A young player suddenly explodes at a major, beats two top seeds, then vanishes in the next round. I can measure what happened in that tournament, but I cannot measure his true physical state, changes in his coaching team, or undisclosed psychological pressure. My current data does not see those things. And the most honest thing I can do is say that I do not see them.

There is another aspect of data that I regard as the darkest side effect of the digitization of sport. Live data supplied to betting companies has turned every point of the ball into a trading signal. A server receiving data a few fractions of a second faster than a rival can gain an edge before the audience even sees the ball land. Technology makes the match more transparent to viewers, but it also opens an information layer accessible only to a small group. That is another data gap — not between those who know and those who do not, but between those who know first and those who know later.

Signals for the next round

From that night of the empty sheet in Brisbane, I drew a new discipline that I apply to every tennis analysis. Before writing any conclusion, I check three questions. Is my data sufficient for the question I am asking? Is my sample large enough that the confidence interval does not render the conclusion meaningless? And what in the current data can I not see?

Those three questions do not make my writing more certain. They make it more honest. And in an industry where everyone wants a decisive answer, honesty is a long-term competitive advantage.

For the rest of this major season, I will track three signals. First, the break-point save rate of top players when facing seeded opponents — this is the metric least polluted by weaker opposition. Second, the gap between overall points-won rate and points-won rate in decisive games, because that gap shows who wins through structure and who wins through moments. Third, recovery speed after long five-set matches, measured by serving quality in the first game of the next match.

These three signals share one trait. They are all measurable, all capable of being wrong, and all capable of being refuted. That is exactly what I want in a signal. A signal that cannot be refuted is not a signal — it is a belief dressed in numbers.

I still keep the habit from when I was sixteen: build a new tracking sheet for every round, and record the empty cells too. Because an empty cell, if recorded honestly, is not a failure of analysis. It is part of analysis. A good analyst is not someone who always has the answer. It is someone who knows exactly where he does not yet have an answer, and says so before someone else finds out.

The next match starts in a few days. My sheet will fill up again. But I will not forget that night in Brisbane, when every cell returned N/A and I almost wrote a story instead of a conclusion. The difference between those two things is my entire trade. And if there is one thing I want readers to carry away from this piece, it is this: the next time you read a tennis analysis full of numbers, ask yourself which part of it is evidence, and which part is only a gap painted over very beautifully.