One target. It strafes, it jukes, it will not hold still. Your job is simple and brutal: keep the crosshair on it. Not near it. On it. That is a completely different skill from the snap-flick your reaction time test hints at — flicking is one explosive movement, tracking is a continuous fight to stay glued down. Here, measuring it means comparing sampled cursor and target positions, not observing every millisecond of your hand movement.
Practice following a moving target
Lock-On is a visual pointer drill. Review its tracking controls and setup before starting.
Set up Lock-OnWant a simple-response baseline instead? Take the visual reaction test. It measures a different task; a lower reaction-time score does not establish better tracking.
Tracking is closed-loop control
Following a moving target is continuous control — the same family of skill that pilots and surgeons get tested on. The classic laboratory version is the pursuit or critical-tracking task, where the target follows an unpredictable path and the apparatus records how well you stay locked to it (Jex, McDonnell & Phatak, 1966). A tracking game does the same thing, with the target's path generated from a seed so the score can be re-checked against exactly the path you saw. Instead of one reaction, you produce a few thousand tiny corrections, and the interesting information is in how well they add up.
What tracking actually measures
We sample your cursor and the target position densely through the round and compare them moment to moment:
| Signal | What it tells you |
|---|---|
| On-target samples (%) | The fraction of recorded cursor samples inside the target |
| RMS tracking error | Root-mean-square cursor-to-target distance across recorded samples |
| Tracking lag | Estimated cursor-to-target offset from cross-correlation |
The headline number here is on-target samples (%), from 0 to 100. Often described as time-on-target, it is a sample-count ratio, not a time-weighted duration. Uneven sampling can affect it, so keep the setup and protocol fixed when comparing sessions.
How to read your score
On-target samples answer "how many recorded positions stayed inside the target?" The other two help describe "how am I losing it?" A larger RMS error means larger sampled cursor-to-target distances. Tracking lag estimates the offset that best aligns the recorded cursor and seed-derived target paths; it does not diagnose whether you were reacting or predicting. Read changes beside on-target samples and sample rate, rather than assuming one metric must improve first.
There is no validated "good tracking is X%" band for these games yet, so we won't print one. Your score is your setup's score: compare it to your own trend and to like-for-like hardware, and let the in-app card show the exact figure. If you want a broader training framework around drills like this, see reaction time training drills; for where raw reaction speed fits in the aiming picture, read reaction time vs aim speed. The flick side of aiming — and its throughput metric — is covered in aim throughput and Fitts's law.
Like every aim metric here, this is the browser-reported pointer in screen space, not your hand and not a motion tracker. Sensitivity, DPI, and OS acceleration all move the numbers, so time-on-target is same-setup relative — a personal-best and equal-hardware metric, not an absolute human ranking.
The honest limits
Pointer polling and browser capture cadence limit what is recorded. Lock-On reports an estimated
sample rate and a cross-correlation estimate of tracking lag, rounded to whole milliseconds for
display. That rounding is not a claim of millisecond accuracy or of precision matched to the
sampling interval. Lock-On does not provide a metric confidence interval or widen one automatically
when samples are sparse. A training game also stacks more hardware into the measurement than the
simple test does (mouse, sensitivity, monitor), which is why its scores stay relative. We only see
the browser-observed stimulus and your event.timeStamp response — never photons, nerves, or muscles.
This is a training range, not a lab tracker.
Ready to inspect your tracking samples? Open Lock-On setup, then choose when to start. Treat a change as a result on that task, not proof of transfer to a game.
Sources & context
The critical/pursuit tracking paradigm behind this game is Jex, McDonnell & Phatak, 1966; that work used dedicated control hardware, not a mouse in a browser, so treat it as the model behind the metric rather than a fixed benchmark for your setup. The cursor and its timing come from the browser's pointer events (MDN documents PointerEvent and Event.timeStamp). For how WiredReflex timestamps browser-observed events and explains timing limits, read how we measure. Those timing limits do not establish confidence intervals for on-target samples, tracking error or lag.