Compare 2–4 players through results, serve/return and shot volumes — with the sample behind each figure.
The period applies to surface records, serve/return and shot volumes. Annual trends, head-to-head and the cumulative profile have their own stated windows. Missing data are not zero.
Recorded historical meetings, independent of the selected period. Not a prediction.
Win percentages grouped by surface.
| Surface | Madison Keys | Shuai Zhang |
|---|---|---|
| Hard | 62%220 won / 355 matches | 50.4%200 won / 397 matches |
| Clay | 65.6%101 won / 154 matches | 36.1%44 won / 122 matches |
| Grass | 77.2%61 won / 79 matches | 53.1%34 won / 64 matches |
Matches without a recognised surface are not assigned to a surface.
Annual win rate across all surfaces; current year incomplete. Independent of the period selector.
Three score-derived indicators, across surfaces and levels. Each needs at least 20 relevant events. This profile does not follow the period selector.
| Metric | Madison Keys | Shuai Zhang |
|---|---|---|
| How they win and how they loseDerived from the score — covers every circuit, Challenger and ITF included. | ||
| Comes back after losing the 1st set | 33.8%n=77 | 11%n=100 |
| Closes out after winning the 1st set | 85.4%n=157 | 88.6%n=105 |
| Wins the tiebreak | 62.1%n=66 | 50%n=40 |
Five-year cumulative score profile, independent of the period selector. Comebacks count matches after losing set one; closeouts count matches after winning it; tiebreaks count tiebreaks. Each rate needs at least 20 events. “—” means unavailable or insufficient data, never zero.
Six observed rates, with opportunities and valid-match samples visible below.
Only valid numerator–denominator pairs from the same match contribute to each rate. Break conversion is derived from the opponent's break points faced minus those saved. The table shows the contributing counts and matches.
| Metric | Madison Keys | Shuai Zhang |
|---|---|---|
| 1st Serve Win % | 68.2%3,111 / 4,564 opportunitiesn=95 matches | 63.6%2,929 / 4,602 opportunitiesn=104 matches |
| 2nd Serve Win % | 47%1,149 / 2,445 opportunitiesn=95 matches | 45.8%965 / 2,105 opportunitiesn=103 matches |
| 1st Return Won % | 38.5%532 / 1,382 opportunitiesn=27 matches | 33.8%284 / 839 opportunitiesn=22 matches |
| 2nd Return Won % | 53.4%407 / 762 opportunitiesn=27 matches | 56.9%330 / 580 opportunitiesn=22 matches |
| BP Saved % | 58.5%295 / 504 opportunitiesn=70 matches | 55.1%313 / 568 opportunitiesn=78 matches |
| BP Converted % | 44.2%277 / 627 opportunitiesn=70 matches | 49%315 / 643 opportunitiesn=80 matches |
Winners, aces, net points won and unforced errors per match. A description, not a skill score.
Not enough common axes to draw a radar. At least three metrics need five recorded matches for every selected player and the reference. Available figures remain below.
Index 100 = recorded circuit average in the selected period, pooling levels, surfaces and match formats. A larger area means more recorded shots or errors per match, not a better player.
Circuit reference: 20.6 · n=1,188
Circuit reference: 1.7 · n=69,314
Circuit reference: 8.1 · n=1,154
Circuit reference: 27.5 · n=1,190
Five matches is a plotting threshold, not a reliability guarantee. Each metric has its own sample. Counts per match also depend on match length.
These are observations from our match database, not an independent verification of every source record. Coverage varies by player and metric; some serve counts are reconstructed upstream from rounded percentages. Surface, opponent strength and match format are not adjusted for. Small samples and different schedules limit comparisons. Career means the history available in our database, not necessarily a complete career.
The weighted head-to-head percentage has been removed: its weights were not validated. Stored deciding-set, serve and partial-set indicators with calculation or coverage issues are withheld here pending upstream correction. Descriptive comparisons do not demonstrate predictive skill.