Sensitivity Finder

What changes between rounds?

We generate five values at −10%, −5%, 0%, +5% and +10% of your starting sensitivity, rounded to six decimals. The order is shuffled. While you test, only the requested sensitivity is shown; offsets are revealed at the end. Seeing the numeric value means the experiment is not fully blind.

The baseline gain is a browser calibration you choose, not a conversion from Valorant yaw. Each candidate has 15 targets: five moves each of 140, 240 and 340 logical units, with fresh directions and shuffled distances. Arena dimensions, target size, gain calibration and input mode stay fixed.

How is the best tested value chosen?

The same Flick Score is used for every candidate: 100 × accuracy fraction² × 1000 ÷ average acquisition milliseconds. Miss rate is 100% minus accuracy. We do not rank by speed alone or claim distance-normalized physiological efficiency.

Highest unrounded score wins. Numerical ties prefer the candidate closest to your starting value, then the smaller value. One short round per candidate is noisy; randomization reduces predictable ordering, but cannot eliminate learning or fatigue.

Why relative mouse input?

With Pointer Lock, the browser hides the OS cursor and supplies movement deltas. We move a virtual crosshair, so different gains genuinely change the same input stream’s displacement. The crosshair is bounded by the virtual arena; the OS cursor’s screen edges do not limit movement.

Is this raw input?

You can request unadjusted movement during preparation. Some browsers reject or ignore that option; we label it as requested, not hardware-verified. Standard relative input can include OS acceleration. MEDQE Tools never silently changes modes mid-experiment.

What does this tell me about Valorant?

Only which relative gain performed best in this 2D task under your chosen calibration. There is no 3D camera, projection or weapon behavior here. Validate a candidate manually in-game and repeat on different days before making large changes.

What stays on my device?

The last 10 completed experiments, including settings and candidate summaries, stay in local storage. No mouse stream is uploaded. No telemetry, accounts or cloud synchronization are used. Clear History removes these saved experiments with confirmation.