A Reading That Changes Meaning Depending on Which Unit You Used
October 24, 2026
Two measurements can share a name, a notation and a scale, and still describe different things. When that happens the numbers stop being comparable, and nothing in the notation warns you. A tennis dataset shows the cost precisely: two readings of the same phrase differ by fourteen percentage points.
The phrase is “ahead in the second set”. It can mean the set has been won, levelling the match. It can also mean the games are ahead inside a set still being played.
The Two Readings, Measured
The corpus covers 116,382 completed best-of-three singles matches from January 2023 to September 2026. For each it records the position of the player who lost the first set, and whether that player won.
Take the second set 6-4 and the match is level at one set each. The win probability from there is 50.16%, across 8,132 matches. Take it 6-1 and it is 58.53% over 3,682. Take it 6-2, 56.38%. Take it 6-3, 52.19%.
Now the other reading. Leading 3-0 in games inside the second set gives 45.22% across 8,755 matches. Leading 3-1 gives 39.02% over 18,412. Leading 2-0 gives 36.8% over 16,660.
So “ahead in the second” lands anywhere between 36.8% and 58.53%, depending only on the unit meant.
Why the Gap Is So Wide
A set score is a completed result. A game score is a position still in progress. Winning the second set converts a deficit into parity, and parity is close to a coin toss. Leading by games has converted nothing, and the set can still be lost.
Written down the two look identical. “6-4” and “3-0” are both two numbers with a hyphen between them. One is finished, one is a snapshot, and the notation does not distinguish them.
The direction is invisible too. Trailing 6-5 in the second set reads 31.93% across 11,369 matches. That is lower than any game-score position above, because the player is about to lose the set.
What This Costs in a Record
Confusion like this is harmless in conversation, because the context is visible to everyone watching. It becomes expensive once the figure is in a column and the context has gone.
Three habits fix it.
Put the Unit in the Field Name
A column called second_set_game_score cannot be read as a set score, and costs eight characters.
Publish the Counts Beside the Rates
The game-score positions here carry 8,755, 18,412 and 16,660 observations. The set-score positions carry 8,132 and 3,682. Different weight, visible at a glance.
State the Counting Rule
In this dataset each position is counted once per match, and the source says so. Without that, a match passing through several positions could be counted several times.
The Two Questions to Ask
What is the number a percentage of, and what unit is the position measured in? Here the answers are matches, and either sets or games. Both are needed before the figure means anything at all.
All of the above comes 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 figures are published at blog.livetennisapi.com/studies.json under CC BY 4.0, free to reuse with attribution.
RJ 21 Which City? Nagaur RTO Code, District Name and Vehicle Registration Details