No game publishes how fast it turns, so the simulator needs one number from you. In a private match or firing range, face a landmark, push the right stick fully right and time how long one full turn takes. Do 3 turns and divide by 3. Do it once more while aiming down sights.
Six short tests, about two minutes. They measure what this page receives from your controller.
Right stick aims. LT aims down sights. RT fires. A starts. Menu pauses, View restarts. D-pad up/down picks a drill, left/right the difficulty. Speed:
Advice uses only runs with your current settings, and waits until there is enough data.
A quick measured check, the usual reasons, and where your game's settings are.
Every setting in your game's menu, in menu order, with the value to use and why.
What decides fights with each class, which drills train it and which setting matters.
Settings on the controller itself. They stack with the game's settings.
Run the benchmark, change one setting, run it again: the settings with the higher average win.
Best and recent scores per drill with your current settings.
Measured means it came from your controller or your own timing. Estimated means Stick Lab calculated it from a model. Community means it comes from guides or patch notes and may be out of date. Only the MW4 menu was checked against real screenshots; other games come from official pages, patch notes and guides.
The browser receives stick values after Windows, the driver and any remapper such as Steam Input or the Xbox Accessories app. Stick Lab measures drift (resting offset), noise, release offset, the smallest movement reported, reach and roundness (5 degree bins), effective resolution (distinct values seen, a lower bound) and triggers. It cannot measure true polling rate or input latency. The report rate shown is what this page receives, not what the controller sends. Grades: drift up to 1% excellent, 1 to 5% normal, 5 to 10% watch it, over 10% problem. Roundness error up to 5% excellent, up to 8% good. Other testers use slightly different cutoffs.
You stand still in a closed range and every target appears in front of you, so no drill turns into a 360. Look speed comes from your timed 360. Without it, speed is a guess and is labelled Estimated. Deadzones are scaled radial unless the game is known to differ. Curves follow each game's own description (for example MW4 Standard starts slow, Dynamic starts fast), but no game publishes its formula, so the shape is approximate. Look boost is added from the game's delay and ramp plus a second timed 360 with boost on.
The aim assist picker shows your game's own options, such as MW3's four Aim Assist Types or Fortnite's strength slider. Each option drives a model with two parts: slowdown near a target, and rotational follow while you move a stick. The strengths are Stick Lab's estimates from patch notes and player tests, because no game publishes its formula. Assisted runs are scored apart and never used for settings advice or leaderboards.
Every run is recorded 30 times per second: where you looked, whether you aimed down sights, every shot and where each target was. The replay draws your path on top. Green means the crosshair was on a target, yellow close, red far. The dashed line is the straight path from where you were when the target appeared. The last 25 runs and your best run per drill and difficulty are kept in this browser.
Perfect: hit with one clean move, no overshoot, first move within 15% of the target. Overflick: the crosshair went past the target edge. Underflick: the first move stopped more than 15% short. Fixed: hit after two or more corrections. Missed: the target timed out. Hit confirm: time from settling on the target to pressing RT.
The coach is a set of fixed rules, not an AI. Each note comes from one measured number (overshoot rate, reaction, lag, wobble, your average) and points at a drill or a setting. The training plan needs 10 runs from the last 14 days.
Six drills at Normal with aim assist off and the same target order every time. Each drill gives up to 1000 skill points. Tiers run Iron, Bronze, Silver, Gold, Platinum, Diamond and Master, each with divisions III to I, then Grandmaster. Your tier is held back by your weakest drill. A drill's own tier is what you would reach if every drill scored like it. The point scale and tier cut-offs are Stick Lab's own and provisional until there is real player data. Differences under about 5% between runs are normal noise.
Each profile on this device keeps its own runs, settings and progress in this browser. An online account on the Stick Lab server adds leaderboards against other players, the real tier distribution and copying your progress to another device. Only scores, dates and your chosen name are sent; your password is stored as a salted hash. Scores are sent by the browser, so leaderboards trust the client.
Endpoint error: where your first fast move stops, compared with the target distance (the first move ends when speed drops below 20% of its peak). Positive is past the target. Overshoot: the crosshair went further than the target edge. Corrections: direction changes after the first move. Settle time: time until the crosshair stays on the target for 120 ms. Lag: how far behind a moving target your crosshair sits, along its direction of travel. Wobble: direction reversals per second while near the target. Second by second: hits (or share of time on target) for each second of the run.
Sensitivity advice needs 30 or more flicks with the same settings and uses a 95% bootstrap confidence range. Stick Lab only suggests a change when the whole range sits outside a normal band (15% short to 10% past the target), changes at most 15% per step, and waits if your times are still improving fast. ADS advice needs 3 tracking runs while aiming down sights. Deadzone advice uses your measured drift plus three times the noise, or the worst release offset, plus 2%. These rules are Stick Lab's own. No study yet shows that settings found in a simulator transfer exactly to a game, so always confirm in game.
Full research report: controller testing methods, all testers and trainers, settings and aim assist for 13 games, aim skill science. Key sources include Gamepadla, the W3C Gamepad spec, Microsoft XInput docs, Rogers et al. 2024 on aim trainer reliability, Natapov et al. 2009 on thumbstick pointing, MacKenzie 2001 on path accuracy, Blizzard's Season 20 aim assist post and each game's patch notes.