The Tennis Desk
Tennis Desk Research · USTA NorCal · Adult league

The Desk level picks line winners more often than USTA's running rating

The Desk level is the Desk's measure of how well a player plays now. In 51,409 USTA NorCal doubles lines of 2025, the side with the higher Desk level won 74.3% of the time. USTA's running rating is the number that USTA uses to set year-end NTRP ratings. The side with the higher running rating won 70.9% of the time.

One limit

Each figure describes many lines, so neither number can say who wins one line.

How do the two numbers compare?

The Desk level was better in doubles and in singles. Each cell is the share of lines that the side with the higher number won.

LineHigher Desk level wonHigher running rating won
Doubles74.3%70.9%
Singles75.0%72.7%

Which number is better to compare before a match?

The Desk level. It is the better guide to who wins a line. Each player page shows the Desk level, and search finds a player page by name. The running rating decides who moves at the year-end, so the year-end forecast uses it.

Who is on the team you play next?
Look up any player or team →

All Tennis Desk Research·Browse every league the Desk holds, by year and area

How this was measured

Scope. The Desk used 51,409 doubles lines and 11,528 singles lines of 2025. The lines come from the USTA NorCal leagues that the Desk covers, in all areas, age groups and NTRP ratings. Each player in these lines has a Desk level, a running rating and an NTRP rating.

The running rating. USTA does not publish it. So the Desk calculates it from the same scorecards that USTA uses.

The rules. Each number is the number that the player held before the line. If both sides had the same number, the Desk gave that pick half credit. The Desk left out defaults, lines that USTA reversed, and lines of fewer than six games.

Measured. The Desk checked every figure above against the live database on November 17.

Method. The Desk produced every figure above with the SQL in docs/research/records/forecast-findings.json, and ran it twice to the same result before it built this page.