What makes a big point in tennis: win probability and leverage

Read the match from the score alone. Find the three points that decided it.

Key facts

  1. ~10.5%

    The average break point on tour moves match-win probability by about 10.5 %.

    Sackmann, "Measuring the Impact of Break Points", Heavy Topspin

  2. 02

    Tennis points aren't strictly independent, but the deviations are small enough that treating them as independent is a workable approximation.

    Klaassen & Magnus (2001), Journal of the American Statistical Association

  3. 03

    Updating the serve numbers during a match improves prediction accuracy over a model that holds each player's ability fixed.

    Kovalchik & Reid (2019), International Journal of Forecasting

  4. 04

    Leverage is the difference between a player's win probability after winning a point and after losing it.

Why it works

Tennis scoring is a ladder. Points make games, games make sets, and sets make the match. At each rung, the next point either climbs or falls. If you know how often each player wins a point on serve, you can compute the chance of climbing from any score. That's the whole model.

The chance changes after every point. How much it changes is leverage, and leverage is what people mean by a big point. A point at 5–5 weighs more than a point at 3–0 because more of the match hangs on it.

Leverage isn't volume. A set point at 5–2, 40–0, gets the crowd loud but moves the match little. A break point at 4–4 in the first set is quiet and matters far more. Most matches are decided by three or four points.

The line to say on court

“Every point counts one. They don't all weigh one.”

Drill: Call the changeover

45 min · alone
  1. Watch one set of any live match. At each changeover, call the leader's chance of winning the set, as a percentage, from the score alone, then check it against a win-probability calculator straight away.
  2. Before each game, name the one point in it that would matter most if it arrives.
  3. Count the changeovers where you were within ten points of the model.
Number to beat5 of 7 changeover calls within ten points of the model

More drills like this: watch like an analyst

Faults and fixes

Calling the loud point the big one
Set points at 5–2 are loud but low-leverage.
FixAsk what happens if it's lost before you decide it's big.
Reading a run of points as momentum
Runs happen in shuffled random sequences too.
FixCompare with a random sequence before you call it momentum.
Trusting a single win-probability line
The inputs are estimates, so the line has a band.
FixRead the band and act on the direction, not the decimal.

Questions players ask

How much does a break point matter in tennis?

The average break point on tour moves match-win probability by about 10.5 %. Break points late in a deciding set move it far more.

How is tennis win probability calculated?

From the score and each player's chance of winning a point on serve, working up the ladder of points, games, sets and match. Points are treated as roughly independent.

Is momentum real in tennis?

It's contested. Runs of points also appear in random sequences, so a win-probability line describes the current odds but doesn't prove momentum.

What should I do on a big point?

Know that it's a big point, and play your percentage pattern, not the highlight.

What this doesn't cover

A win-probability line describes the current odds. It doesn't predict. For historical matches the serve numbers are estimates, so the line should be read as a band, not a precise value. Momentum is contested, and this model doesn't claim it.

Sources

  • Kovalchik & Reid (2019), International Journal of Forecasting
  • Klaassen & Magnus (2001), Journal of the American Statistical Association
  • Sackmann, "Measuring the Impact of Break Points", Heavy Topspin
  • 2024 MCM Problem C point-by-point data (2023 Wimbledon)

Read next