RallyIQ

Game plan

Arnaud Clement v Michael Chang

Every number combines what Arnaud Clement does well with what Michael Chang allows, each measured against the tour average and shrunk toward it when the sample is thin. Flip perspective

Forecast

Arnaud Clement wins, best of 3 62%90%: 40%–80% · best of 5: 64%
Serve points won 61.8% / 59.5% Arnaud / Michael · tour 63.8%
Strengths only, no similarity priors 54%serve 60.8% / 60.0%

Each player's serve and return strength is fitted against every opponent they were charted against, so a record built on weak opponents counts for less. At least one of them is no longer active or has too little charted in the last three seasons, so both are compared on their careers. The result is then nudged by Arnaud Clement's record against Michael Chang's tactical lookalikes and in their charted head-to-heads (lookalikes: +7.5 on serve, +4.2 on return vs expectation (454 points)). A game-by-game Markov chain turns point odds into match odds; the 90% range covers the uncertainty in the two strengths, not the nudges. Charted matches lean toward big events, so treat this as a scouting estimate, not a betting line.

Head to head, per 100 shots

CareerArnaudMichael
Direction choice+0.22 ±0.12
better than 95%
−0.01 ±0.13
better than 54%
Shot selection+0.48 ±0.40
better than 91%
−0.02 ±0.19
better than 49%
Execution−0.52 ±0.57
better than 44%
+1.06 ±0.36
better than 96%
Points left on the table2.21 ±0.13
lower than 91%
2.88 ±0.27
lower than 26%

Each player's career against an average tour player in the same position, adjusted for opponent strength, with a 90% margin (shots clustered by match). Points left on the table is the gap to the best-value direction for the same stroke, so lower is better. Percentiles are within each player's own tour. A side is highlighted only when the gap is larger than the margin on the difference.

Serve plan

The share of points the server wins when a first serve lands in that direction (hover a rate for its 90% interval; ± is the 90% margin). "Matchup" combines the server's rate with how this returner handles that serve. "Optimal" is the mix that wins most against this returner once they start reading a habit, at the response measured across the tour, and only within the range servers' habits actually vary. The gain over the current mix is how exploitable that mix is.

Arnaud Clement serving

Deuce court

1st serveNowArnaud winsv MichaelMatchupOptimal
Wide50%70%75%72.6%±5.664% ▲
Body6%65%59%61.1%±12.10% ▼
T43%72%79%76.1%±6.036% ▼

Optimal v Michael Chang: +0.5±0.7 per 100 first serves (faults included) over the current mix, inside the 90% margin. Serving T every time would read +2.7 per 100 first serves in before the returner adjusts.

Ad court

1st serveNowArnaud winsv MichaelMatchupOptimal
Wide51%65%67%58.8%±6.738% ▼
Body5%54%62%53.8%±13.05%
T44%66%74%68.4%±6.957% ▲

Optimal v Michael Chang: +1.0±0.8 per 100 first serves (faults included) over the current mix. Serving T every time would read +5.7 per 100 first serves in before the returner adjusts.

Michael Chang serving

Deuce court

1st serveNowMichael winsv ArnaudMatchupOptimal
Wide41%65%68%59.0%±6.837% ▼
Body8%56%60%52.2%±13.60% ▼
T50%69%78%72.8%±6.663% ▲

Optimal v Arnaud Clement: +0.4±0.7 per 100 first serves (faults included) over the current mix, inside the 90% margin. Serving T every time would read +7.4 per 100 first serves in before the returner adjusts.

Ad court

1st serveNowMichael winsv ArnaudMatchupOptimal
Wide48%65%75%68.2%±6.761% ▲
Body7%59%59%55.1%±13.30% ▼
T44%63%76%68.5%±7.339% ▼

Optimal v Arnaud Clement: +0.2±0.7 per 100 first serves (faults included) over the current mix, inside the 90% margin. Serving T every time would read +1.2 per 100 first serves in before the returner adjusts.

Return plan

Value of each return, in points per 100 returns against an average return of the same serve (direction, court, surface): the tour's result with that return, the returner's own edge with it, and what this server gives up when it comes back to that side. Returns with no charted direction are left out, so values compare with each other rather than with zero. Depth isn't a choice here: missed returns have no depth. Serve quality isn't charted, so a block through the middle partly reflects the serve that forced it.

Arnaud Clement returning

1st serve to the forehand

ReturnNowTourOwnv MichaelValue
FH through the middle38%+4.3+2.3+1.2+7.8±2.5
FH down the line28%+1.7+1.6+2.7+6.0±4.0
FH crosscourt17%+5.5−3.1+4.4+6.8±4.0
FH slice through the middle11%−4.2+0.8+2.7−0.7±2.3
FH slice down the line4%−4.3+2.6+0.9−0.7±2.5

Lean FH through the middle: +2.2±2.1 per 100 returns v the current mix (256 returns charted)

1st serve to the backhand

ReturnNowTourOwnv MichaelValue
BH crosscourt46%+8.8+0.4+2.4+11.6±3.2
BH through the middle28%+6.4−1.2−0.1+5.1±2.6
BH down the line11%+4.2−0.4+2.0+5.8±4.5
BH slice through the middle8%−4.2−2.0+5.0−1.2±2.1
BH slice crosscourt4%+0.5−0.7+5.7+5.5±2.2

Lean BH crosscourt: +4.3±1.9 per 100 returns v the current mix (236 returns charted)

2nd serve to the forehand

ReturnNowTourOwnv MichaelValue
FH down the line37%−1.8−0.9+2.4−0.3±4.1
FH through the middle35%−3.5+1.7+2.5+0.8±2.6
FH crosscourt28%−0.2−1.5+1.6−0.1±3.8

Lean FH through the middle: +0.6±2.5 per 100 returns v the current mix (100 returns charted, inside the 90% margin)

2nd serve to the backhand

ReturnNowTourOwnv MichaelValue
BH crosscourt44%+1.0+0.9+0.4+2.3±3.0
BH down the line29%−0.2−0.6+0.4−0.4±5.0
BH through the middle27%−2.9−0.3+2.2−1.0±2.4

Lean BH crosscourt: +1.7±2.3 per 100 returns v the current mix (177 returns charted, inside the 90% margin)

Michael Chang returning

1st serve to the forehand

ReturnNowTourOwnv ArnaudValue
FH through the middle53%+4.3+1.7+1.3+7.3±2.4
FH down the line29%+1.7+2.6−3.3+1.0±4.1
FH crosscourt17%+5.5−2.6−2.3+0.6±4.3
BH through the middle1%+6.6+0.4+0.3+7.2±2.3

Lean FH through the middle: +3.0±1.8 per 100 returns v the current mix (676 returns charted)

1st serve to the backhand

ReturnNowTourOwnv ArnaudValue
BH through the middle44%+6.4−2.4+0.3+4.3±2.4
BH crosscourt32%+8.8+5.1−0.3+13.6±3.1
BH down the line12%+4.2+4.0+1.8+10.0±4.5
BH slice through the middle6%−4.2−0.9−1.1−6.2±2.5
BH slice crosscourt4%+0.5+2.1+2.5+5.1±3.0

Lean BH crosscourt: +6.5±2.4 per 100 returns v the current mix (718 returns charted)

2nd serve to the forehand

ReturnNowTourOwnv ArnaudValue
FH through the middle47%−3.5+1.7+1.0−0.8±2.7
FH down the line26%−1.8+0.9+1.4+0.5±4.7
FH crosscourt26%−0.2+1.4+1.7+2.8±4.4
BH inside-in2%+2.5−0.4+0.9+3.0±2.9

Lean FH crosscourt: +2.3±3.7 per 100 returns v the current mix (286 returns charted, inside the 90% margin)

2nd serve to the backhand

ReturnNowTourOwnv ArnaudValue
BH crosscourt50%+1.0+1.5+0.9+3.4±2.9
BH through the middle35%−2.9+1.0+1.3−0.7±2.3
BH down the line12%−0.2+3.9−1.3+2.4±5.0
FH through the middle2%−2.9−1.0+1.0−2.9±2.1
FH inside-out1%+1.0+0.4+1.4+2.7±3.1

Lean BH crosscourt: +1.7±1.7 per 100 returns v the current mix (486 returns charted, inside the 90% margin)

Rally plan

Edge, in points per 100 shots: the hitter's skill with the shot (own) plus how much the receiver usually gives up against it (theirs), both measured against the tour average on hard. Each player's hard record is shrunk toward their all-surface one, so a thin sample on it moves the numbers only a little.

Arnaud Clement

Favour

ShotEdgeOwnTheirs
FH to their backhand · serve +1+6.6±3.8+1.4+5.2
FH to the middle · return+4.5±2.9+3.1+1.4
BH to their backhand · return+3.6±3.4+1.3+2.3
FH to their backhand · rally+3.0±2.5+0.2+2.9
FH to their backhand · return+2.4±4.3+1.4+1.0
FH to their forehand · return +1+2.1±4.9−0.7+2.8

Avoid

ShotEdgeOwnTheirs
BH to their forehand · return +1−3.1±6.7−0.4−2.8
FH to their forehand · rally−2.5±2.6−3.6+1.1
FH to their forehand · return−2.4±5.2−4.9+2.5
FH to the middle · rally−1.9±2.5−1.1−0.8
BH to their forehand · rally−1.7±4.3−0.7−1.0

Michael Chang

Favour

ShotEdgeOwnTheirs
BH to their forehand · return+8.3±5.7+6.4+1.9
BH to their backhand · return+5.4±3.3+3.5+1.9
FH to their backhand · serve +1+4.8±4.0+2.7+2.1
BH to the middle · serve +1+3.4±3.3+1.0+2.4
FH to their forehand · rally+3.1±2.7+2.8+0.3
FH to their backhand · rally+2.8±2.8+1.5+1.3

Avoid

ShotEdgeOwnTheirs
FH to their forehand · return−5.6±5.5−2.6−3.0
BH to the middle · return +1−1.9±3.4−0.7−1.1
FH to their backhand · return−1.5±4.5+1.2−2.7
FH to their forehand · return +1−1.5±5.1+3.7−5.2
FH to the middle · return +1−1.4±3.9−2.0+0.6

Against Michael Chang-like opponents

Arnaud Clement vMatchesServe pts wonReturn pts won
All charted opponents–58.3%38.5%
Players most similar to Michael Chang2 65.7%41.1%

Similar by tactical fingerprint: Andy Murray, Marcos Baghdatis, Lleyton Hewitt, Andy Roddick, David Nalbandian, Marat Safin, Juan Carlos Ferrero, Magnus Norman, Aaron Krickstein. When two players have rarely met, their records against these lookalikes fill the gap.