The Empty Spreadsheet in Tennis: When Historical Data Series Break

Three in the morning in New York, I opened a spreadsheet and found it empty.

Not empty because I had forgotten to save it. Empty because everything I needed to put into it — first-serve percentage, points won on second serve, break-point conversion rate, winner-to-unforced-error ratio — exists in no public data source I am permitted to access. On the second monitor, a match was still running. A player called his coach to the court at the eleventh game of the third set. The commentator talked about competitive character. The broadcast graphic showed exactly one thing: the score.

That feeling comes back every time a match moves into a zone the historical data can no longer describe. Fans watch with their eyes; I watch with a probability distribution — but this time the probability distribution had nothing to distribute.

Tennis is among the most densely measured individual sports there is. Every ATP or WTA event emits point-by-point data. Electronic line-calling systems have replaced most line judges at the bigger tournaments, which means every serve leaves behind a coordinate. The 25-second shot clock, adopted at ATP level from 2026, turns time into a measurable variable. The rule allowing coaches to come onto court, made permanent by the ATP from 2026, opened an entirely new data layer.

Yet those same changes are what cut the series apart. A player with a 68% win rate on second serve across 2026-2026 no longer carries the same meaning once off-court coaching alters the psychological structure of deciding games. The old denominator is not wrong. It simply no longer measures the same thing.

Based on my experience watching matches, most analytical errors do not come from reading a number incorrectly. They come from reading a number collected under the old rules and applying it to a match played under the new ones.

When the spreadsheet is empty, there are four layers of checking I have to pass through before I am allowed to say anything at all.

The first layer is declaring the gap. It sounds pointless, but the sentence "I have no sample" is worth more than a conclusion that happens to be right by accident. In my spreadsheets, empty cells carry a dedicated marker; I never leave them blank, because a blank cell is far too easy to fill with a guess when you are tired.

The second layer is mapping the break. Every time a rule changes — shot clock, off-court coaching, electronic line calling, seeding regulations — I draw a vertical line on the time axis. Any comparison crossing that line has to be annotated. The rule break always matters more than the numbers sitting on either side of it.

The third layer is rebuilding a baseline. If there is no sample under current conditions, I downgrade the question: instead of asking how often this player converts break points, I ask how this player responds when trailing in a set. A narrower question can be answered through structured observation, and it still retains predictive value.

The fourth layer is assigning probability instead of conclusion. There is no "certain", no "deserving". Only "roughly 60% likely", "narrow confidence band", "sample insufficient to separate signal from noise".

These four layers do not make me smarter. They only make me harder to catch out.

There is one detail the media rarely mentions: most tennis data is published as seasonal aggregates, not by rule condition. A leading player such as Novak Djokovic, Carlos Alcaraz or Jannik Sinner can be cited with a single career-long percentage, when that percentage is stitched together from at least three different rule eras. The number does not lie. The person quoting it is the one telling a story that does not exist.

At Grand Slam level the pressure is greater still. US Open 2026 total prize money was announced by the organisers at roughly 75 million USD — a fact that is easy to quote, easy to headline, and tells you nothing whatsoever about the quality of play. Every number in a contract is a confession by the market. But it confesses about the market, not about tennis.

The awkward part is that the tennis industry produces its most abundant storytelling precisely when the data is thinnest.

A coach is appointed, and the press release contains not one metric. A young player reaches a semifinal, and is immediately labelled the next generation, even though the sample is five matches. A player returns from injury on a protected ranking, and the rankings list becomes an administrative document nobody reads closely.

Correlation is not causation — the phrase has been repeated so often it has become a cliché, yet it remains true in an uncomfortable way. When data is thick, stories have to crowd into a narrow space. When data is empty, the space opens up, and anything at all can be called a cause.

I have made exactly that mistake. After a major tournament, I used a single basic indicator to conclude that a team did not deserve to advance, and I was asked to review the entire tape. The result was a month of retreat into research. Since then, every piece I write carries a mandatory section at the end: data limitations. The market forgets nothing; it merely disguises itself as a new summer. The truth sits deep beneath the table of numbers, where headlines never reach.

The signal worth tracking next round is not which player is winning the most. It is which tournament will start publishing data by rule condition instead of by season. Whoever does that first will hold an information edge that is hard to copy. Everyone else is still reading the score and calling it the truth.