Three drills, three failure modes
Aim trainers generally split aiming into clicking, target switching and tracking, because each breaks in its own way and practising one does little for the others. Precision removes the clock: with unlimited time, whatever error is left is placement rather than haste. Flicks puts every target far from the last one and asks for speed, which is where the two classic failures appear — the flick that sails past because the hand never stopped, and the flick that stops short because it started braking too early. Tracking asks you to stay on a moving target, where the failure is not distance but delay.
Overshoot and undershoot
In the flick drill each click is measured along the line you travelled, not only as a distance. Landing beyond the centre in the direction of travel is overshoot; landing before it is undershoot. The two want opposite corrections, which is exactly what a single accuracy percentage hides. Overshooting means the movement needs a firmer stop, and lowering sensitivity a little usually helps; undershooting means the flick is braked too early, and moving slightly faster with one decisive stop tends to fix more than slowing down. The plot is rotated so travel always points right — a cloud sitting right of centre is overshoot you can see.
Lead and lag
Tracking is scored as time on target: the share of the run your cursor spent inside the target while you held the button. The diagnostic number is the second one. Your offset is projected onto the direction the target is moving, which separates being behind it — the ordinary case, since you are reacting to a change of direction you could not know was coming — from being ahead of it, which means you are anticipating and overcorrecting. Lag responds to following the target rather than chasing it; lead responds to trusting its current direction less.
What the numbers are worth
All distances are CSS pixels in this window, so a run on a trackpad and a run on a 27-inch screen are not comparable, and neither are two runs at very different window sizes. There is no leaderboard and no percentile here: the only honest comparison is against your own earlier runs in the same drill at the same size, which is what the history strip shows. This is a diagnostic rather than a training programme — it tells you which way you are wrong, and a dedicated trainer is the place to drill it out.
Built as part of gptgames.dev .