Conditions get the credit for almost everything. A pitch is heavy, a court is quick, the air is thin, and the result follows. It is a satisfying explanation because you can see conditions and you cannot see much else.
Professional tennis is a good place to test that. Conditions there change more than in most sports, and the record of what happened is unusually complete. Across 2,239,440 service games played between January 2023 and September 2026, here is what the ground is worth.
A server holds 74.43% of the time on grass, 69.95% on hard courts and 65.98% on clay. Grass to clay is a gap of 8.45 percentage points.
That is a real effect and it is not a small one. Now put it next to the other thing you could have measured.
The gap that is more than twice as wide
Sort the same games by which circuit they were played on and the spread opens up.
Men’s top-tour matches hold 78.62% across 230,715 games. The men’s Challenger tier holds 74.1% over 622,018, and the men’s ITF circuit 70.26% over 529,948. On the women’s side the top tour holds 64.36% across 266,503 games. The Challenger tier holds 60.4% over 92,396 and the ITF circuit 58.93% over 553,157.
Top to bottom that is 19.69 percentage points, against 8.45 for the surface. Who is on court moves the number more than twice as much as what they are standing on.
Nobody finds this surprising once it is written down. It is still the opposite of how the sport is usually discussed. Surfaces have names and seasons and arguments attached to them. The depth of a field has none of that.
Where the ground does show up
Change the question and the surface comes straight back.
Count how often a set goes to a tiebreak and grass gives 15.91%, hard courts 12.31% and clay 10.72%. Per match, 31.73% of grass matches contain at least one tiebreak, against 25.28% on hard and 22.52% on clay. A grass match is roughly 40% more likely to produce a set nobody could separate.
So the ground is specific. It decides how close sets get, and barely touches who wins the service game itself.
Why this keeps happening outside tennis
The pattern repeats anywhere a number is explained after the fact. The visible variable takes the credit, and the one that moves the result most is the one nobody photographs.
A shop blames the weather for a quiet Tuesday. The weather is real and it does something. The staffing rota probably did more, and the rota is not interesting to look at.
The test is cheap. Before accepting any explanation for a difference, ask what else you could have grouped the data by. Group it that way and compare the two spreads. If the explanation you started with produces the smaller gap, treat it as a minor contributor and go looking for the bigger one.
That takes one afternoon on data you already hold, and it needs no statistics beyond subtraction.
What a usable version of this looks like
Three habits are enough.
Publish the denominator next to every rate, the way 230,715 sits next to 78.62%. A percentage with no count behind it cannot be argued with or checked.
Report more than one grouping even when one of them is boring. The clay figure is only meaningful because the ITF figure is printed beside it.
Say which groupings you tried. A report showing a single cut has either tested one idea or picked the best-looking of several. The reader cannot tell which.
None of that makes the first explanation wrong. It puts a size on it, and the size is the part that usually goes missing.
The figures above come from the Live Tennis API, a JSON service for developers. It covers live scores, fixtures, players and match events, with point-by-point history behind them. The data is published openly under CC BY 4.0, free to reuse with attribution.
Method: completed best-of-three singles matches on the ATP, WTA, Challenger and ITF circuits, 1 January 2023 to 14 September 2026. Retirements and walkovers excluded. Tiebreak games are excluded from the hold rates. Counted from the point-by-point record, not estimated.
Also Read-Everyday Elegance: What It’s Like Living in a Lucerne Grand Home at Thomson Reserve