Match Point — workshop

One-input tennis rally mechanic — swing timing alone decides the shot, dead-on-time going back safe and central, early or late peeling off toward a line.

A tennis mechanic with a single input and no aim stick: when you swing relative to the ball’s arrival is the entire decision. Swing on the ideal instant and the return goes back safe and central, easy to read, hard to miss. Swing early or late and the same swing peels the ball off toward one line or the other — further off-time, more angle, more pace, and more chance it’s a clean winner the rally can’t answer. Push past a cutoff on either side and there’s no shot at all, just a mistimed swing into the net. Every rally speeds up the longer it runs, so the safe, central window gets less forgiving in real time even though it never moves in the abstract.

Timed to the US Open, whose qualifying rounds are underway in Flushing this week ahead of the main draw on August 30.

Reuse

The mechanic lives in src/match-point.js as a framework-free ES module:

createRally(opts) — returns a state object; call update(dt) each frame and swing() on input. No aim or power parameters — the only decision is when swing() is called relative to the ball’s arrival.
opts.rand overrides the RNG (unused by the core resolution, reserved for future opponent variance).
State exposes phase (incoming | outgoing | result), t (the timing clock — 1 is the ideal contact instant), ball (x, y, court-normalized), cfg.cleanWindow/cfg.missWindow (the safe and total-cutoff |t - 1| radii, for drawing a timing gauge), lastOutcome ({ won, reason }, reason one of winner | rally-cap | mishit | missed), and score (player/ai/winners/unforced/rally/bestRally/streak/bestStreak). Rendering, input, and any timing-gauge UI are entirely the caller’s job — the demo drives it with spacebar or click and draws the gauge itself.

Port notes for DragonRuby: update() and resolveShot() have no DOM dependencies; the court-to-screen mapping and the timing gauge belong to the renderer, same split as the demo.

Gotchas

swing() is a no-op outside phase === 'incoming' — the module advances itself through outgoing and result on its own, no caller-side polling needed beyond calling update(dt) every frame.
Direction and pace both derive from the same offset = t - 1; there’s no separate aim axis. A shot inside cleanWindow always has angle === 0 — the risk zone (between cleanWindow and missWindow) is where angle and pace both ramp up together, which is what makes going for a line a strictly riskier decision than staying central rather than just a cosmetic one.
rallyPace (and with it the ball’s real-time speed) grows every exchange within a point and resets only when a new point starts — a mid-rally screenshot will show a visibly faster gauge sweep than the serve.
The demo in demo/ bundles its own copy of the module (self-contained by contract). If you touch src/, re-copy it into demo/.