How to disguise your serve and hit the T, body and wide

Hit three spots from one toss. Give less away.

Traced from real footage

See it, eight instants at a time.

Every frame is traced from real match footage, so you see the shape of the stroke, not a drawing of an idea.

Federer’s serve, instant 6 of 8
FIG. 27 · FEDERER’S SERVE
Contact
Federer’s serve: eight instants traced from the footage, frame for frame. Hover or tap to flip; the dot marks contact.

Key facts

  1. 01

    Eight current or former male tour players described combining serve-motion cues with match context. This interview study did not measure a universal prediction time.

    Vernon, Farrow & Reid (2018)

  2. 02

    In six right-handed professionals, first-serve foot, toss-apex and impact positions differed by target; second-serve kinematics were consistent across targets.

    Reid, Whiteside & Elliott (2011), author abstract

  3. 03

    A matching toss does not establish that every other cue is hidden. The lesson removes two authored cues to demonstrate the idea; real anticipation needs its own test.

    Lesson model limitation

  4. 04

    Treat each target as an intention: T can restrict return angles, a body serve can crowd the swing, and wide can move the returner. Position and execution determine what actually comes back.

  5. 05

    This lesson sets a practice target of 7 successful serves out of 10 attempts to each zone. It is an authored aim, not a validated disguise score.

  6. 2025

    ATP reported Andy Roddick’s assessment that Jakub Mensik could reach all four corners from the same toss after Miami 2025. This is an expert observation, not a measured toss-dispersion study.

    Eichenholz, ATP Tour (22 April 2025), reporting Roddick’s assessment

  7. 2025

    In his June 2025 interview with Roddick, Agassi discusses Becker’s tongue cue and says Becker probably still denies the habit. This is a retrospective player account, not a validated cue-accuracy result.

    The Andre Agassi Interview, Served with Andy Roddick (3 June 2025), 49:40–51:46

Why it works

Choose the target with the next ball in mind. T may limit the returner’s angles; a body serve aims to reduce swing room; wide may move them and open space. Watch the actual reply and the returner’s recovery before choosing your next shot. A landing lane alone does not guarantee any of these effects.

Inspect whether your preparation changes with your intended target. The model exaggerates toss and racquet marks so you can see a tell. In a match, the returner can also use other movements and your previous choices. Keep watching the ball after contact; removing a tell does not remove the need to return the actual serve.

Keep several executable options under pressure. An equal split across T, body and wide is an illustration, not a required distribution. Weigh your reliable serves, the returner’s position and the reply you want; three recent winners alone do not settle the next choice.

The line to say on court

“Same toss, three answers.”

Drill: Same toss, three corners

15 min · with a partner
  1. Mark broad T, middle and wide landing lanes inside the deuce service box with flat markers. Warm up, agree the boundaries and choose one familiar serve type for this block. Start without a returner.
  2. Choose the target privately before the toss. Use a comfortable, repeatable preparation; do not change direction in response to a late call.
  3. Hit ten cycles of three serves, one to each lane per cycle in a varied order: 30 attempts, ten per lane. Faults and wrong-lane serves stay in the count. Rest as needed; if you stop early, a partial round doesn't count toward the number to beat.
  4. Count legal landings in the chosen lane. Keep all three counts; after ten attempts per lane, your number is the lowest of the three. Then play ten separate live points with a returner. Count legal landings on the intended target, and watch the actual reply separately. Check whether your middle-lane serve crowds this returner; do not add these live serves to the lane score.
Number to beat7 of 10 legal landings in each marked lane

More drills like this: serve

Faults and fixes

Your toss moves with the target
A target-linked change may give the returner useful information.
FixChoose the target before the toss and compare your preparation across targets. Practise a repeatable motion you can execute comfortably; do not chase a late target call.
Chasing the line
Aiming too close to a line can leave little room for your usual error.
FixInclude body serves and choose a target you can reach reliably.
The same spot on every big point
Being predictable under pressure is exploitable.
FixChart your choices and results, then consider other serves you can execute.

Questions players ask

Where should I serve in tennis?

Choose among the targets you can execute, considering the returner’s position and the next ball you want. T, body and wide offer different possibilities, but an equal distribution is not required. Check what the return actually does.

How do returners read the serve?

Returners can combine the server’s movement with previous patterns and the situation. No single prediction time or pair of cues describes every returner and server.

Is placement more important than speed on the serve?

We have no evidence here for a speed threshold above which placement or disguise matters more. Work on the combination you can execute and check the resulting returns.

Should I keep serving to the spot that's working?

It can remain a good choice. Weigh how well you can execute it, the returner’s adjustments and your alternatives; you do not have to force an equal target distribution.

What this doesn't cover

The model’s toss shifts and face marks are authored. A matching toss cannot prove that a real serve is unreadable. The 7-of-10 drill scores target execution, not anticipation or disguise. There is no measured speed threshold or required target distribution here. The Mensik example is an expert assessment; the Becker example is Agassi’s retrospective account. Neither measures cue accuracy or a causal effect on results. The ten cycles and ten live points are authored practice blocks, not validated training doses. Calibrated film measurements and measured transfer remain open.

Sources

Read next