Tennis winning streaks: do players on a run win more?

A player who has just won five in a row is “in form”: it is the stock phrase. But does that player really win more than their level suggests? In other words, does a winning streak carry information that the Elo rating does not already have?

We took 38,071 ATP and WTA singles matches, played between January 2022 and September 2026, and looked, for each player, at how many wins they had strung together before walking on court. Then we compared their result with what their Elo predicted. The answer is clear, and it runs against intuition: the player on a streak wins less than expected.

Five wins in a row, three points below expected

At equal Elo probability, we measure the gap between what the player wins and what they should win, by the length of their streak. Each match counts twice, once for each player.

The numbers move in one direction and grow with the streak: −1.5 points after 3 or 4 wins, −3.3 points after 5 to 7, −3.8 points beyond. Taken together, players on 5 wins in a row or more win 56.9% of their matches while Elo expected 60.3%: a gap of −3.4 points, with a 95% interval from −4.8 to −1.9 (3,714 observations, z = −4.6). A gap this size is very unlikely to come from chance.

It is not a quirk of the sample

To check that the gap does not come from a detail of the sample, we cut it other ways. These splits were made after seeing the total: they are checks, not discoveries.

• It exists on both tours. −3.9 points on the ATP Tour (1,876 observations), −2.8 points on the WTA (1,838).
• It affects favourites and underdogs alike. A player on a streak who is the favourite wins 68.2% against 70.7% expected (−2.5 points). As the underdog, 30.9% against 36.3% (−5.4 points, 1,122 observations).
• It does not come from an opponent who is also on a streak. Against an opponent with no marked streak (between two losses and two wins in a row), the player on a streak wins 58.2% against 62.3% expected: −4.2 points. Removing the 266 matches where both players had five wins in a row, we get −4.0 points.

A gap that shrinks over time

By year, the gap for players on 5 wins or more is −4.0 points in 2022, −5.8 in 2023 and −4.2 in 2024, then −1.0 in 2025 and −1.8 in 2026 (up to 22 September). The last two years can no longer be told from chance, with about 750 and 670 observations each. It may be an effect fading, or plain sampling noise: our data cannot settle it. What it does show is that you cannot count on a stable three-point gap.

What about losing streaks?

You might expect the opposite, a player losing confidence and doing worse than predicted. That is not what we see: after 5 losses in a row or more, the player wins 39.1% of their matches against 37.4% expected, +1.7 points, with a 95% interval from −0.5 to +3.8 that contains zero. Losing streaks add nothing measurable. Only winning streaks drift from Elo, and in the direction of disappointment.

Bookmakers are less wrong

On matches where we have the bookmakers’ odds, about one match in two, the player on 5 wins or more wins 56.2% of their matches. Odds without margin gave 56.8%, a gap of −0.6 point. On the same matches, Elo alone gave 58.8%: −2.6 points, from −4.6 to −0.6.

So the market is less wrong than Elo on these players, but the sample is too small to say it is not wrong at all: the interval of its gap runs from −2.6 to +1.4 points, and −2.6 is inside it. Either way, a gap of a few points is not enough to make money against the bookmakers’ margin: our betting guide explains why.

How to read a winning streak

• A streak does not add to the probability. A player on 6 wins is not stronger than their Elo says: at equal probability, they win a little less often than expected.
• Do not turn it into a contrarian signal either. The gap is three points on matches where Elo expects 60%. The player on a streak still wins 56.9% of their matches, and the gap has shrunk over the last two years.
• Look at what really changes: fatigue and rest days, the surface or the head-to-head.
• This figure is about Elo alone, not the full Pronocast model, which relies on more than the rating (see our methodology).

Today’s predictions show each player’s 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, in which both players had already played at least twenty matches: 38,071 matches, seen from both sides. A player’s streak is the number of consecutive wins (or losses) before the match, across all tours, ITF and Challenger included, computed by replaying matches in order: it is read before the match and updated after, with no leakage of the match’s own result. The expected rate is the actual win rate observed for a given Elo probability, in 2.5-point bands, over the whole sample. Gaps are judged with a z-test, and the 95% interval is the gap plus or minus 1.96 times its standard error. The five-win threshold is the one used to read the result; every streak band is shown. A single match can appear in two groups when both players have a streak. Odds are the bookmaker average, converted to a probability without margin, available for 18,749 matches of the sample. The full script is kept with our data.