Best of five in tennis: do favourites really win more?

In a best-of-five match, is the favourite better protected? It is the classic argument for the Grand Slam men’s format: to beat a better player you have to outplay them three times, not two, and luck has fewer openings. The reasoning holds on paper. But it is hard to check, because every best-of-five match is played in the same place: at a Grand Slam, against the best players in the world.

We went back over 33,794 ATP and WTA singles matches played between January 2022 and September 2026, 2,405 of them best-of-five, and compared each result with what both players’ Elo ratings predicted. To separate the effect of the format from the effect of the Grand Slam, we have a rare control group: the women, who play the same tournaments on the same courts, but over three sets.

What the theory says

If a player wins every set with the same probability, independently of the others, the calculation is direct. The longer the match, the more likely the better player is to win.

A typical favourite in our sample, at 67% over three sets, would therefore win about 4 points more over five. That is the figure to keep in mind: the theory predicts a real but modest advantage.

At the Slams, favourites win more than expected

To find out whether the favourite wins more than they should, we compare their win rate with the one Elo leads us to expect, calculated only on best-of-three matches.

70.8% against 63.9%: on the ATP tour, the favourite wins seven points more at the Slams than elsewhere, and 3.4 points more than Elo expected, a gap that does not happen by chance (z = 3.65). At this stage, everything seems to vindicate the format.

But the women do the same over three sets

Look at the second row. In the women’s Grand Slams, the favourite wins 70.5% of their matches, 2.8 points more than expected. It is all there, with only three sets. The difference between the two tours, once the expected value is removed, is +0.6 points, with a 95% interval from −2.0 to +3.1 points. We cannot say that five sets make the favourite win more: what we are measuring first is a Grand Slam effect.

That effect does not depend on the surface. Surface by surface, the picture is the same on hard courts, clay and grass.

Why do the Slams favour favourites so much? We have not tested it. The usual suspects are maximum motivation for the best players, a 128-player draw with protected seeds, and players who prepare specifically for these events. Those are hypotheses, not results.

A hint in the later rounds

The five sets are not entirely absent from the numbers. Split by round, a difference between men and women does appear, but not where you might expect it.

In the first two rounds, the men do no better than the women. From the third round on, the ATP favourite wins 77.3% of their matches, 8.7 points more than expected, while the WTA favourite stays at the expected level. The difference is 7.0 points, with a 95% interval from +1.8 to +12.1. It exceeds the 4 points the theory predicts: a reason to stay cautious rather than to conclude.

Read it carefully: we split by round after seeing the numbers, which raises the risk of finding a gap that does not exist, and the sample is modest (581 ATP matches). It is a hint that five sets weigh in the second half of the draw, not yet proof.

The bookmakers already know

On matches where we have the bookmakers’ odds, the market favourite wins 73.5% of their matches at the men’s Grand Slams (1,323 matches), while margin-free odds gave them 72.1%. The 1.3-point gap cannot be told apart from chance. For comparison, the odds give the favourite 66.0% at other ATP events: the market has already priced in the Grand Slam premium, and did not overstate it. At the women’s Slams, it announces 71.0% for 73.3% achieved.

So there is no obvious loophole: if you think “the favourite is worth more over five sets”, the odds think so too. Our betting guide explains why an idea like that is not enough to make money.

How to read a best-of-five match

• Don’t treat the favourite as unbeatable. At the men’s Grand Slams, they still lose almost three matches in ten.
• The first round is no safer than elsewhere. Our data show no five-set-specific advantage there.
• If the format matters, it is in the later rounds. From the third round to the final, the ATP favourite wins more than expected, which the WTA favourite does not.
• Look at fatigue too. A five-set match wears players out more than three sets, and our fatigue study shows what rest days change, or don’t.

In Pronocast, the format matters for the set score: the model knows whether a match is played over three or five sets, and its five-set score table is measured on more than 2,800 matches. For the win probability, it receives neither the format nor the tournament, but it does receive the odds, which already include it. The predictions of the day show that score next to the probability, and the verified results page shows what they are worth once the matches are played.

Method

ATP and WTA singles matches from 1 January 2022 to 22 September 2026, where both players had already played at least twenty matches: 33,794 matches. Best-of-five matches are those of the men’s main draw at the Australian Open, Roland-Garros, Wimbledon and the US Open (2,405 matches). Qualifying matches, played over three sets, are removed: our data provider labels them as quarter-finals, semi-finals or finals, so we keep the matches played from the first first-round match of each edition. A few matches are missing from the archive: 108 to 127 matches per edition and tour, against 127 in a complete draw. The expected rate is the actual win rate observed for a given Elo probability, in 2.5-point bands, calculated on best-of-three matches only. Gaps are judged with a z test, and the interval for the difference between two groups is that difference plus or minus 1.96 times its standard error. Odds are the bookmakers’ average, converted to a margin-free probability, available for 16,469 matches in the sample. The full script is kept with our data.