ATP · Right-handed · 4 charted matches · 1991–1997
Paul Haarhuis
Against an average opponent
Serve and return points won, refitted against every opponent at once so a record built on weak or strong opposition is put on the same scale. Career, all surfaces, with a 90% margin. Raw is the plain share of points won.
Value per 100 shots
Points gained per 100 shots compared with an average tour player in the same position, adjusted for the strength of the opponents faced, with a 90% margin (shots clustered by match). Raw is before the opponent adjustment. Built on 2,024 shots.
Shot expected value
The share of points Paul Haarhuis goes on to win after each option in the positions they face most often, shrunk toward tour average when the sample is small. Showing the 8 most-used options; teal marks the best one with at least 30 shots.
Rally, shots 5–8: drive to your backhand side
position worth 46% to the average player · 71 shots
| Option | Used | Win % | Tour |
|---|---|---|---|
| BH crosscourt | 42% | 41.0%±11.4 | 47.6% |
| BH through the middle | 41% | 36.2%±11.3 | 43.7% |
Rally, shots 5–8: drive to your forehand side
position worth 44% to the average player · 52 shots
| Option | Used | Win % | Tour |
|---|---|---|---|
| FH crosscourt | 42% | 46.0%±12.7 | 46.6% |
| FH through the middle | 29% | 46.6%±13.9 | 41.5% |
| FH down the line | 29% | 34.1%±13.2 | 44.7% |
Serve +1: drive to your middle, you at net
position worth 61% to the average player · 52 shots
| Option | Used | Win % | Tour |
|---|---|---|---|
| FH volley down the line | 33% | 61.5%±13.2 | 63.8% |
| FH volley crosscourt | 25% | 68.9%±13.3 | 68.7% |
| BH volley crosscourt | 23% | 64.3%±13.9 | 62.9% |
Rally, shots 5–8: drive to your middle
position worth 51% to the average player · 44 shots
| Option | Used | Win % | Tour |
|---|---|---|---|
| FH crosscourt | 32% | 54.5%±14.0 | 52.7% |
| FH down the line | 23% | 51.0%±15.0 | 51.5% |
Serve under pressure
Pressure predictability index ±0 How much less varied Paul Haarhuis's first-serve direction gets on break points. Positive means easier to read. Based on 49 break-point first serves.
Deuce court
| 1st serve | Usage | Break pt | Won when in |
|---|---|---|---|
| Wide | 41% | 50% ▲ | 73% / 73% |
| Body | 18% | 8% ▼ | 52% / 63% |
| T | 40% | 42% | 74% / 75% |
235 normal · 12 break-point 1st serves
Ad court
| 1st serve | Usage | Break pt | Won when in |
|---|---|---|---|
| Wide | 46% | 49% | 70% / 73% |
| Body | 13% | 19% | 64% / 63% |
| T | 41% | 32% ▼ | 71% / 72% |
201 normal · 37 break-point 1st serves
Is the serve mix in equilibrium?
Game theory says a well-mixed server wins equally often with every direction they use. If one direction wins more, it's underused and points are being left behind. This is the minimax test Walker and Wooders ran on Wimbledon finals, applied to every charted first serve. Win rates include faults. "Optimal" allows for returners reading a habit. No measurable response (−0.05 ± 0.15 points per 100 serves for every 10 points of habitual usage), measured from ATP servers whose mix drifted between matches. Shifts stay within the range servers' habits actually vary, the only range that response was measured over.
Deuce court
| 1st serve | Usage | Points won | Optimal |
|---|---|---|---|
| Wide | 42% | 66.4%±6.7 n=103 | 55% ▲ |
| Body | 18% | 50.2%±9.6 n=44 | 5% ▼ |
| T | 40% | 54.7%±7.2 n=100 | 40% |
Off equilibrium (p = 0.011): serve wide more. Gap 7.6 points per 100 first serves.
Optimal mix: +1.1 per 100 first serves.
Ad court
| 1st serve | Usage | Points won | Optimal |
|---|---|---|---|
| Wide | 47% | 60.9%±6.8 n=111 | 46% |
| Body | 14% | 65.6%±9.8 n=33 | 1% ▼ |
| T | 40% | 66.4%±7.0 n=94 | 53% ▲ |
Consistent with an optimal mix (p = 0.52).
Optimal mix: +0.7 per 100 first serves.
Exploitability 0.87 points per 100 first serves What the optimal mix would win over the current one, both courts. More exploitable than 100% of ATP servers. Tested on matches they weren't fitted on, ATP mixes picked this way win 0.33 per 100 first serves on average.
Repeating the previous direction to the same court: +0.9±5.3 points per 100 against switching. Negative means returners read repeats. Tour-wide, repeating costs women about 0.4 points per 100 and costs men nothing, so men's returners don't measurably anticipate direction. (216 repeats, 261 switches.)
Return by serve direction
Return points won against each serve direction, compared with the tour average.
| Serve | Court | Direction | Points | Won | vs tour | |
|---|---|---|---|---|---|---|
| 1st | Ad court | Body | 14 | 37% | −0.4±11.9 | |
| 1st | Ad court | T | 50 | 24% | −3.8±7.9 | |
| 1st | Ad court | Wide | 54 | 23% | −4.4±7.5 | |
| 1st | Deuce court | Body | 19 | 33% | −4.0±11.0 | |
| 1st | Deuce court | T | 68 | 18% | −7.2±6.4 | |
| 1st | Deuce court | Wide | 59 | 23% | −4.5±7.3 | |
| 2nd | Ad court | Body | 31 | 42% | −7.1±10.4 | |
| 2nd | Ad court | T | 18 | 52% | +2.4±11.9 | |
| 2nd | Ad court | Wide | 42 | 42% | −6.0±9.6 | |
| 2nd | Deuce court | Body | 27 | 47% | −2.2±10.9 | |
| 2nd | Deuce court | T | 37 | 52% | +2.4±10.0 | |
| 2nd | Deuce court | Wide | 25 | 41% | −7.3±10.9 |
Signature patterns
Recurring sequences that win more than Paul Haarhuis's own baseline, ranked by edge weighted by how often they're used.
Serve → +1
- Not enough data
Return
- vs T serve (ad court) → FH through the middle used 13.0% · won 22% · +1.9±9.8 vs own baseline
Rally, consecutive own shots
- Not enough data
Discovered sequences
Mined from every me → opponent → me run of three shots, with no templates. Ranked by how much more often Paul Haarhuis wins the point once the sequence happens, weighted by how often it happens. Think of them as chess openings.
- BH crosscourt → FH crosscourt → BH through the middle used 2.0% · won 36% · −7.3±12.4 vs own baseline · −20.9 vs tour on the same sequence Disrupted by Jimmy Connors (4/16)
Strengths and vulnerabilities
Value per 100 shots compared with the average player hitting (strengths) or facing (vulnerabilities) the same shot. Only shot types seen at least 120 times.
Hurts opponents most with
| BH to the middle · return | +0.9 | 120 |