Break point conversion is a statistic that mostly measures luck

A player faces so few of them in a season that the percentage swings wildly for reasons that have nothing to do with nerve.

Somebody converted three of eleven and the discussion writes itself. Fragile under pressure. Cannot close. The mental side. Then the same player goes seven from nine a week later and the story becomes redemption, and nobody notices that both of those samples are far too small to support any story at all.

Run the arithmetic in your head. A tour player might see a couple of hundred break points across a full season of singles. Split by surface, by opponent quality, by best of three against best of five, and the meaningful buckets have a few dozen points in them. At those counts a conversion rate can move ten points either side of a player's true ability through nothing but the ordinary lumpiness of small numbers. Broadcast graphics present it to two decimal places anyway.

I am not arguing that pressure is imaginary. It plainly is not. I am arguing that this particular statistic is a very noisy instrument for detecting it, and that we reach for it because it is available rather than because it is informative.

If you genuinely want to see whether somebody's game changes under load, look at behaviour rather than outcome. Does the first serve percentage drop on break points, and does the second serve get slower and shorter. Does the rally shot go a metre higher over the net and a metre shorter into the court, which is the classic signature of a player playing not to lose. Does the time between points stretch. Those are measurable, they occur on every pressure point regardless of whether the point was won, and so they accumulate sample far faster than a conversion percentage ever can.

There is also a selection problem nobody mentions. Break points against a great server are much harder to convert than break points against a weak one, and the players who reach the deep rounds spend their break points against the former. Punishing them for a low conversion rate is close to punishing them for playing good opponents.

The cleanest fix is one the tours could implement tomorrow. Report conversion against expectation, given the server faced. It exists in the data already.

Until then, the number on screen is a coin sequence with a narrator attached.