Players who moved up a level won 50.7% of their lines the next season, and 9.7% moved back down
- Lines won in the first season up
- 50.7%
- 4,937 lines by 693 players
- Moved back down a year later
- 9.7%
- 60 of 621 players
- Played a line at the new rating
- 60.2%
- 693 of 1,151 players, against 59.8% for players who stayed at their level
Players who moved up a level won about half their lines in their first season. In Marin and San Francisco, players moved up 1,151 times from 2022 to 2025, and after 693 of those moves they played at the new rating the next season. They won 50.7% of the 4,937 rated lines they played, where a line is one of the matches inside a team match. Most of them stayed up, though a year later about one in ten had gone back down a level: 9.7% of the 621 still rated there.
These are past seasons for groups of players, so no figure here is a forecast for any one player.
How the first season went
On the forums, the worry is a first season without a single win. It rarely happened. Only 10 of the 352 who played 6 or more lines at the new rating won none (2.8%). Most of them, 196 players, won at least half their lines (55.7%).
Players new to 4.0
Players who moved up from 3.5 to 4.0 won a little under half their lines: the 246 who played at 4.0 won 48.2% of their 1,886. Players already at 4.0 won more often in the same seasons, 54.7% of their 16,222 lines. Most of the newcomers stayed up all the same. A year later, about one in eleven of the 198 still rated had gone back down (9.1%).
Before and after the move
The 461 players who played at both ratings won 79.1% of their rated lines at the old rating the season before, but 52.3% at the new one. Some will read the drop in wins after a move up as a sign that the new level was too much. I see it differently: the rating did its job. Players who had won most of their lines at their old level won about half at the new one.
For these same players, the drop got bigger at each step up the ratings. The 83 who moved from 4.0 to 4.5 fell furthest. They had won 80.9% of their lines at 4.0, but 46.7% at 4.5.
Beside the players already there
The players who moved up won less often than the players they joined, who won 54.1% of their 45,591 lines in the same seasons. They won less at three of four steps, but more at 3.0 to 3.5. The gap was widest for the 59 who moved from 2.5 to 3.0: they won 39.0% of their 344 lines, against 54.2% for the players already at 3.0.
| Players who moved up | Already at the new rating | ||||
|---|---|---|---|---|---|
| Step | Players | Lines | Won | Lines | Won |
| 2.5 to 3.0 | 59 | 344 | 39.0% | 6,071 | 54.2% |
| 3.0 to 3.5 | 242 | 2,032 | 57.4% | 15,383 | 53.4% |
| 3.5 to 4.0 | 246 | 1,886 | 48.2% | 16,222 | 54.7% |
| 4.0 to 4.5 | 106 | 555 | 45.6% | 6,858 | 53.4% |
| 4.5 to 5.0 | 40 | 120 | too few | 1,056 | 58.4% |
"Too few" marks a group too small to give a rate: fewer than fifty players, or fewer than five moves back down.
Every step, women and men
| Step | Moved up | Played at the new rating | Lines won, first season | Still rated a year later | Moved back down |
|---|---|---|---|---|---|
| All, 2.5 to 3.0 | 103 | 59 | 39.0% | 53 | too few |
| All, 3.0 to 3.5 | 413 | 242 | 57.4% | 226 | 4.0% |
| All, 3.5 to 4.0 | 368 | 246 | 48.2% | 198 | 9.1% |
| All, 4.0 to 4.5 | 190 | 106 | 45.6% | 104 | 12.5% |
| All, 4.5 to 5.0 | 74 | 40 | too few | 37 | too few |
Women and men
| Step | Moved up | Played at the new rating | Lines won, first season | Still rated a year later | Moved back down |
|---|---|---|---|---|---|
| Women, 2.5 to 3.0 | 99 | 55 | 39.0% | 49 | too few |
| Women, 3.0 to 3.5 | 319 | 183 | 59.3% | 181 | 4.4% |
| Women, 3.5 to 4.0 | 236 | 162 | 48.1% | 140 | 7.9% |
| Women, 4.0 to 4.5 | 98 | 60 | 49.1% | 60 | 8.3% |
| Women, 4.5 to 5.0 | 35 | 16 | too few | 20 | too few |
| Men, 2.5 to 3.0 | 4 | 4 | too few | 4 | too few |
| Men, 3.0 to 3.5 | 94 | 59 | 48.6% | 45 | too few |
| Men, 3.5 to 4.0 | 132 | 84 | 48.5% | 58 | 12.1% |
| Men, 4.0 to 4.5 | 92 | 46 | too few | 44 | too few |
| Men, 4.5 to 5.0 | 39 | 24 | too few | 17 | too few |
Moving back down
A move back down was more common after a move up than among players who stayed at their level, who moved down 3.2% of the time. A few went the other way, though, and moved up again: 26 of the 621.
How the first season went made a difference. Players who won at least half of 6 or more lines rarely went back down: fewer than five of 131 did. But those who won under half went back down more often: 15.0% of 107. The 204 who played no rated line at the new rating went back down nearly as often, 12.7% of the time.
About a quarter of the players who moved up from 2022 to 2024 had no published rating in either area a year later, 212 of the 833. The 9.7% leaves out these 212 players, so if they were struggling, the true rate of moving back down is higher. Players who stayed at their level left the records about as often, 26.2% of the time.
Playing on at the new rating
Most players who moved up kept playing at the new rating, 60.2% of the 1,151. Players who stayed at their level played on about as often.
A season without a rated line could mean no team, an injury, or only Combo and Mixed, which are left out because neither is played at one rating.
Free, no signup. The Tennis Desk reads the real USTA NorCal match record.
All Tennis Desk Research·Browse every league the Desk holds, by year and area
How this was measured
Scope. USTA NorCal adult league in Marin County and San Francisco. The moves up came at the 2022 to 2025 year ends, and each one was followed into the rated lines of the season after it.
Who is counted.
- A move up is a computer-rated player's published NTRP rating rising one level at a year end, and there were 1,151 of them.
- A move back down is a rating one level lower again at the next year end. Only the 2022 to 2024 moves are followed, because the 2025 moves meet their next year end in December 2026.
- A player's rating in a season is the one on their team rosters. A season in which their rosters list two different ratings is left out.
- Each person is followed by their USTA player id, so two people with one name are never merged. Each move up counts once, so a player who moved up at two year ends counts twice.
- The chart and the step table count different players. The chart counts only the 461 players who played at both ratings. They are the same people before and after, so the drop is measured in one group. The step table counts everyone who played at the new rating, so its shares differ a little from the chart's. Both of its groups played their lines at the new rating, in the first season after the move.
- Moves back down by rating. Of the 60 who moved back down, 47 held a computer rating and 13 an appealed one, a rating the player asked USTA to review. Of all 621 still rated a year later, 7.6% moved down to a computer rating and 2.1% to an appealed one.
- For players who stayed at their level, each year a player kept their rating counts once, so a player who kept it for three years counts as three years. Their shares count 11,528 years for playing on, 6,121 years for moving down and 8,298 years for leaving the records. Of the 6,121 years, 2.6% ended in a move down to a computer rating and 0.5% in one to an appealed rating.
- Left out: players moving up from a self-rating, because correcting a self-rating is not a bump, and moves of two levels.
- Moves up to an appealed rating or a self-rating count too. 73 of the 1,151 moves up reached an appealed rating and 13 a self-rating. Leaving those out, the players who moved up won 50.8% of 4,413 lines, and 9.9% of 576 moved back down.
Who moves up in the first place is the subject of Where do you have to finish to get bumped?
Where a line counts. A rated line counts in an adult 18 & Over or 40 & Over league named by one rating, daytime and flighted leagues included. Combo, Mixed and 55 & Over leagues are named by the sum of two ratings, so a line there is at no single rating. A league named with a plus, such as 2.5+, is open to two ratings and is left out too. So the chart has no 2.5 row: from 2022 to 2025 the only 2.5 leagues here were 2.5+ leagues.
Measured. Every figure above was checked against the live database on October 5.
Method. A group under fifty players, or with fewer than five moves, shows its size instead of a rate. The SQL behind every figure is recorded in docs/research/records/what-happens-after-a-bump.json, so anyone can re-run it with npx tsx scripts/measure-research-record.ts.