Method
What this measurement rests on, what it reproduces, and where it stops being supported by evidence. This page exists because the interesting part of a sensitivity tool is not the number it gives you — it is whether that number means anything.
The two rules
1. Measure in counts, never in centimetres
Your mouse reports counts. A game converts each count into a rotation through a constant called yaw. So the sensitivity you want is:
sensitivity = 360 / (counts per 360° × yaw)
DPI does not appear. It cancels out, because counts are what the hardware actually sends. Every tool that starts by asking for your DPI inherits whatever error is in the number you typed, and most people's DPI is a rounded figure from a marketing page. This app uses your DPI for exactly one thing — showing you a distance in centimetres — and works without it.
2. Never present a guess as a measurement
The app distinguishes, everywhere and visibly, between what it measured, what it derived, and what somebody assumed:
- Each game carries a confidence level. Where a game's sensitivity scale is widely circulated but unverified against the live game, it says so on the result.
- Controller turn rates are null wherever nobody has measured them. Deliberately null — not missing work. The app measures yours instead of guessing.
- The analysis refuses to answer when there is too little data, rather than fitting a curve to noise, and reports a confidence interval when it does answer.
- Rankings are hidden until a benchmark comes from real player scores. Until then you are compared to your own history, which is a real fact.
The research
The flick half reproduces the target geometry and protocol of:
Boudaoud, Spjut & Kim (2022), Mouse Sensitivity Effects in First-Person Targeting Tasks (arXiv:2203.12050).
Two findings matter here. The first is that there is an optimal band — roughly 0.45 to 1.8 degrees per millimetre, or about 20 to 80 cm per 360° — outside which performance falls off for most people.
The second is the one the product is actually built on: the optimum moves with the difficulty of the shots you take. Small targets far away pull it one way; large targets close pull it the other. That is why the result is broken down by difficulty band rather than given as a single number, and why the per-game advice is weighted by what each game asks of you.
What is ours, and not proven
The study measured static targets only. The tracking half of this app — the moving-target measurement — is our own extension. It is built on the same speed-precision reasoning and validated the same way, against simulated players whose optimum is known by construction. But it is an extrapolation, and every screen that shows it says so.
Tracking is not garnish. A controller stick sets a turn rate rather than moving your view a distance, so you cannot flick with one. A controller recommendation built on the flick task alone rests on the wrong measurement, and the app refuses to pretend otherwise.
What is not settled
Stated here rather than buried, because a measuring instrument that hides its error bars is not a measuring instrument.
| Benchmarks | The rank thresholds for each drill are reasoned guesses, marked as such in the data and in the interface. They need recalibrating against real score distributions, which do not exist until people play it. |
|---|---|
| Two games' scales | Fortnite's circulated yaw and slider range are mutually inconsistent, and Warzone changed its scale in December 2024. Both are flagged in the app rather than quietly used. |
| Controller data | No public turn-rate data exists for any game. The app measures yours; it ships none. |
| Playstyle weighting | The per-band optima are measured. How they are blended into one number per game is reasoned, and labelled heuristic. |
Why the measuring room is bare
Measurements run in a deliberately plain room, not in the range you practise in. The study used an empty background so that nothing could hide a target or pull the eye, and a result is only comparable to that study — or to another player — if the room never changes and never depends on your graphics settings.
Targets themselves are unshaded and immune to fog in both rooms, by construction. No quality preset, lighting change or post-processing effect can alter how a target looks. Two things depend on that: measurements are only comparable between people if a target looks the same to everyone, and a leaderboard is worthless if turning the graphics down makes targets easier to see.
Screenshots on this site are taken on the Balanced graphics preset, which is what the range is meant to look like. The shipped default is Competitive, which turns shadows, fog and colour grading off to protect the frame rate. It looks flatter.