Understanding Joystick Circularity & Radial Deviation: What Outer Stick Traces Actually Measure

When you push a controller thumbstick against its outer plastic rim and rotate it on a diagnostic page, the tool draws a perimeter shape and calculates a percentage called Average Radial Deviation (often referred to on forums as "circularity error"). Because that result is displayed as a percentage, many players assume 0.0% means a flawless thumbstick and a higher number means the stick is defective, drifting, or counterfeit.

In reality, a browser circularity test measures one specific geometric property: how closely the outer coordinates reported to your browser match a mathematical circle of radius 1.000 when the stick is tilted past 40% deflection.

Before reading a circularity percentage, it helps to distinguish outer circularity from stick drift. These two tests focus on different parts of the thumbstick's physical travel:

• Stick Drift (Resting Center Behavior): Drift testing looks at what happens near the center origin (0.000, 0.000) when you take your thumb completely off the stick, or right after the stick snaps back to rest. If the resting coordinate stays outside a game's inner deadzone, the camera or character moves on its own.
• Outer Circularity (Full-Tilt Boundary Geometry): Circularity testing looks at the outer boundary when you push the thumbstick shaft all the way against the circular housing collar.

On JoystickTester.com, the circularity tracker ignores all stick movement at or below 40% tilt (r ≤ 0.400). That means small resting offsets or minor center jitter are excluded from the circularity percentage.

Because the two measurements inspect different zones, a high radial deviation percentage on an outer trace does not prove your controller has stick drift, and a low radial deviation percentage does not guarantee your stick centers cleanly at (0.000, 0.000). At the same time, the two regions are not always mechanically isolated: a severely shifted calibration center, a loose gimbal, or a binding sensor mechanism can affect both resting center coordinates and outer boundary symmetry at once. To check resting behavior specifically, pair your outer trace with our Controller Deadzone Test and our Controller Stick Drift Guide.

A web browser cannot measure raw electrical voltage inside a sensor, measure physical thumbstick tilt angle in degrees directly at the gimbal, or measure physical shaft displacement in millimeters along the housing rim. Instead, the browser reads normalized coordinate numbers exposed by the W3C Gamepad API (navigator.getGamepads()) on each display animation frame (requestAnimationFrame).

When a controller uses the W3C "standard" mapping, the browser reports two floating-point numbers for each thumbstick, nominally scaled from -1.000 to +1.000:

• Left Stick: axes[0] (Horizontal X: -1.000 Left to +1.000 Right) and axes[1] (Vertical Y: -1.000 Up to +1.000 Down).
• Right Stick: axes[2] (Horizontal X) and axes[3] (Vertical Y).

Plain-English Explanation of the 72 Sectors:

Think of the circular radar on the Joystick Tester as a compass face divided into 72 equal slices (sectors), where each slice covers 5 degrees of rotation (72 × 5° = 360°).

1. 1
   Whenever you push the stick past 40% tilt (r > 0.400), the tool checks which of the 72 five-degree slices the stick is pointing into and how far from the center (0.000, 0.000) that coordinate sits.
2. 2
   Inside each 5° slice, the tool remembers only the farthest point your stick reached during the current test.
3. 3
   As you roll the stick all the way around the rim, those 72 farthest points are connected in a ring to form the violet outer trace polygon, and each point is compared against the solid gray reference circle (r = 1.000).

Because the tracker stores the farthest point reached in each 5° slice, the way you move your thumb directly shapes the recorded trace. Follow these five steps on the Joystick Tester for a clean, repeatable sweep:

1. 1
   Connect and Wake Your Physical Controller: Plug in your controller via USB or connect it over Bluetooth, click inside the browser tab, and press any physical button so your device appears in Slot 0–Slot 3. Confirm that the amber "Simulated Input Active" banner is not displayed, as simulated mode generates a synthetic mathematical curve rather than reading physical sensors.
2. 2
   Reset Existing Trace Data: Click "Reset Both Traces" in the top header (or the "Reset Trace" button on an individual stick card) so the sector counter resets to 0 / 72 Sectors (0%).
3. 3
   Tilt the Stick to the Outer Collar: Push the thumbstick smoothly until the shaft touches the circular plastic rim. Use steady, comfortable thumb pressure—pressing hard enough to flex the plastic shell or accidentally click L3/R3 (B10/B11) is unnecessary and can cause uneven contact along the rim.
4. 4
   Rotate Slowly 360° in Both Directions: Keeping the shaft lightly against the outer collar, roll the stick slowly around the entire perimeter for 2 to 3 full turns (one clockwise and one counter-clockwise). Rolling slowly gives the browser's frame loop time to capture at least one sample inside every 5° sector.
5. 5
   Verify Sector Coverage Before Reading Deviation: Check the badge at the top of the 72-Sector Outer Circularity Trace panel and confirm it reads 72 / 72 Sectors (100%) (or close to full coverage) before evaluating Avg Radial Dev. or Max Radial Dev.

Once you complete a sweep on the Joystick Tester, three metrics appear below the radar scope:

1. 1
   Angular Coverage (X / 72 Sectors and %):

   Angular coverage shows how many of the 72 five-degree slices have recorded at least one outer sample (r > 0.400) since your last reset.

   • Plain-English Meaning: Coverage measures whether you completed the test movement around the perimeter, not how accurate your controller is. On our page, if coverage is below 60% (fewer than 44 sectors), the summary box displays "Partial Perimeter Trace" to remind you to keep rotating the stick around the rim.
2. 2
   Average Radial Deviation (Avg Radial Dev.):

   Average radial deviation measures how far, on average, your recorded outer trace sits from the r = 1.000 reference circle across all sampled sectors, expressed as a percentage.

   • Plain-English Meaning: If every recorded sector landed right on the r = 1.000 circle, the average deviation would be 0.0%. In a hypothetical numerical example where every sector reached r = 1.080 (8% outside the circle) or r = 0.920 (8% inside the circle), both cases produce the exact same 8.0% average radial deviation because the calculation averages the unsigned distance |r − 1.000| in either direction.
   • Why Average Deviation Cannot Establish Asymmetry Alone: Because Average Radial Deviation averages |r − 1.000| across all sampled sectors without preserving sign or compass direction, it cannot tell you whether a trace is symmetrical or lopsided. A balanced rounded-square trace and an off-center, skewed trace can report the exact same average percentage. Always inspect the actual 2D outer trace polygon and sector dot colors on the radar:

     • Green dots (|r − 1.000| ≤ 0.060): Within ±6% of the r = 1.000 reference circle.
     • Sky-blue dots (r > 1.060): Extending outside the r = 1.000 circle (common in the four diagonal corners).
     • Amber dots (r < 0.940): Stopping short inside the r = 1.000 circle.
3. 3
   Maximum Radial Deviation (Max Radial Dev.):

   Maximum radial deviation shows the single largest percentage gap from r = 1.000 recorded in any one of the sampled 5° sectors.

   • Plain-English Meaning: This highlights peak corner overshoot or the deepest inward flat spot. For instance, in a hypothetical trace where the four cardinal directions (Up, Down, Left, Right) sit at r = 1.000 (0% difference) while the four diagonal corners extend to r = 1.160, Max Radial Dev. will report 16.0% even though Avg Radial Dev. across all 72 sectors is much lower.

Even though the physical plastic opening around almost every modern thumbstick is circular, different controllers—and even the same controller connected through different software layers—often draw either a tight circle or a rounded square (where the four diagonal corners bulge outward past r = 1.000, plotted up to ±1.25 on our radar).

That outer shape is not determined by one single component. Instead, it reflects how five layers interact between your thumb and the browser:

1. 1
   Two Perpendicular Axis Sensors on a Square Coordinate Grid: A thumbstick measures horizontal (X) and vertical (Y) tilt on two perpendicular axes, each normalized to [-1.000, +1.000]. On a 2D graph, the region where both X and Y can reach ±1.000 forms a square whose corners sit at a radial distance of √(1² + 1²) ≈ 1.414.
2. 2
   Mechanical Gate Geometry: The physical plastic collar (or circular/octagonal faceplate ring) stops the thumbstick shaft mechanically as you rotate it around the perimeter. Manufacturing tolerances, shaft diameter, or aftermarket thumbstick caps change how far the gimbal pivots before touching the rim.
3. 3
   Sensor Response, Calibration Tables & Onboard Processing: Inside the controller, raw sensor readings are mapped against stored center and outer calibration endpoints. Some designs normalize the combined (X, Y) vector radially so the outer boundary stays near r = 1.000 at every angle (producing a circular trace). Other designs scale the X and Y axes more independently so tilting into a diagonal corner pushes both axes closer to (±0.80, ±0.80) or higher before clamping at ±1.000 (producing a rounded-square trace with r > 1.000 in the diagonals).
4. 4
   OS Drivers & Input Wrappers: Operating system drivers (Windows XInput, DirectInput, or macOS GCController) and software translation layers (such as Steam Input or third-party remappers) can apply their own radial or axial deadzones, outer threshold scaling, or custom response curves.
5. 5
   Browser W3C Gamepad API Remapping: Finally, your browser normalizes the driver's reported axis values into the W3C [-1.000, +1.000] floating-point array read by the page.

A common misconception in online controller discussions is that a low radial deviation percentage (for example, a hypothetical 1% or 2% reading) proves a controller has higher-grade sensors or proves it uses Hall Effect or TMR magnetic sensors, while a higher reading (such as a hypothetical 10% or 12% rounded-square trace) indicates inferior hardware.

In practice, circularity percentage alone cannot rank controller quality or identify sensor technology:

• Radial Clamping Can Be Applied to Any Sensor Type: Whether a thumbstick uses carbon-film potentiometers, Hall Effect magnetic sensors, or Tunnel Magnetoresistance (TMR) modules, normalizing (x, y) coordinates onto a unit circle (r = 1.000) is a mathematical scaling choice. A low deviation percentage only shows that the final output coordinates are clamped close to a circle—it does not measure internal sensor resolution, centering repeatability, noise floor, or mechanical durability.
• Game Engines Process Diagonal Input Differently: Game developers use different axis-processing algorithms depending on the game genre. A game engine that evaluates polar vector magnitude (r = √(x² + y²)) reaches full input intensity as soon as r reaches 1.000 (which occurs around (0.707, 0.707) on a unit circle). By contrast, a game or emulator that evaluates horizontal X and vertical Y axes independently against separate per-axis thresholds may respond differently to diagonal tilt depending on how far each individual axis extends in the corners. Neither a circular trace nor a rounded-square trace is universally better for every game.

Because our diagnostic records any movement where r > 0.400, an incomplete or rushed sweep can produce misleading numbers on a properly functioning controller:

• Cutting Across the Circle Leaves Artificial Flat Chords: Suppose you push the stick straight Up (r = 1.000), let it pull inward to r = 0.550 as you move toward Right, and then release it. Because 0.550 is above the 0.400 recording gate, the 5° sectors between Up and Right record 0.550 (45% below 1.000), drawing a flat inward line across the upper-right quadrant and raising both Avg Radial Dev. and Max Radial Dev. Pushing the stick all the way against the outer collar and sweeping slowly through that quadrant immediately overwrites those 0.550 points with the true outer boundary.
• Touching Only a Few Angles Skews the Average: Flicking the stick only to Up, Down, Left, and Right samples only a small fraction of the 72 sectors and leaves the four diagonal corners unsampled. Whenever a trace looks jagged or incomplete, check Angular Coverage first, click "Reset Trace", and perform a slow 2–3 turn rotation along the collar until coverage reaches 72 / 72 Sectors (100%).

Instead of treating the deviation percentage as a pass/fail score, use the 72-sector radar on the Joystick Tester to visually inspect symmetry and full cardinal reach:

1. 1
   Balanced Circle or Balanced Rounded Square: If the polygon reaches near the r = 1.000 circle at -Y (Up), +Y (Down), -X (Left), and +X (Right) and looks symmetrical left-to-right and top-to-bottom (whether circular or rounded-square in the corners), the outer boundary is consistent.
2. 2
   Inward Flat Spots or Restricted Reach (r < 1.000 with Amber Dots): If an entire side or corner of the polygon stays well inside the solid r = 1.000 ring even after you slowly roll the stick firmly against the collar, the browser is receiving undersaturated coordinates in that direction.
3. 3
   Lopsided or Shifted Polygon: If the trace extends far past the circle on one side (for example, +X Right) while falling short inside the circle on the opposite side (-X Left), the axis range or center offset is skewed.

Checking Possible Causes Without Jumping to Hardware Failure Conclusions:

An inward flat spot or asymmetric polygon in the browser shows what coordinates the web page received, but it does not by itself prove a broken internal part. Check these possibilities in order:

• Incomplete Sweep Technique: Re-sweep the affected region slowly while maintaining steady contact against the outer collar to rule out an accidental mid-stick shortcut.
• Browser / OS Mapping or Active Software Profiles: Check whether the amber "Non-Standard HID Mapping (mapping: '')" notice appears on the page, or whether a background remapper (such as Steam Input desktop configuration, DS4Windows, or a vendor utility) is applying a custom response curve or outer deadzone limit.
• Shifted Calibration Bounds: If a controller's stored minimum/maximum axis endpoints are out of alignment—or if replacement thumbstick modules were installed without running a range calibration afterward—one direction may cap below 1.000 until recalibrated.
• Physical Rim Interference: Inspect the thumbstick collar for external obstructions, such as a thick silicone thumb-grip sleeve rubbing against the faceplate edge or a shell that is not seated evenly.

Depending on what you observe on the Joystick Tester, two companion diagnostics on JoystickTester.com can help you verify the next step:

1. 1
   Follow Up Undersaturation or Asymmetry on the Controller Calibration Tool:

   • What It Checks: Step 1 samples a 60-frame hands-off resting baseline (meanX, meanY, and peak-to-peak jitter), and Step 2 records your exact minimum and maximum cardinal reach (Min X, Max X, Min Y, Max Y), total axis span, and directional symmetry. On our calibration page, each cardinal direction is compared against an illustrative ≥ 0.950 (95%) UI threshold ("Full Cardinal Saturation (>=95%)"). Step 3 provides a temporary browser-session normalization preview so you can see whether zero-point subtraction and range rescaling balance the stick's output.
   • Choosing an Appropriate Calibration Utility: Because web browsers operate in a read-only security sandbox and cannot write calibration tables to controller hardware or OS drivers, persistent calibration must be performed outside the browser. As detailed in our How to Calibrate Your Controller Guide, available utilities depend on your specific controller model, firmware, and operating system:

     • Microsoft Xbox Accessories App (Windows & Xbox Consoles): Provides a guided firmware thumbstick recalibration wizard for supported official Xbox Wireless controllers, depending on controller model, firmware version, and feature availability inside the app (check whether "Recalibration options" appears for your specific controller in Xbox Accessories).
     • Windows USB Game Controllers Setup (joy.cpl): Provides a local Windows DirectInput registry axis calibration wizard for standard USB gamepads and joysticks.
     • Sony PlayStation Accessories App (Windows PC): Provides official firmware updates for DualSense wireless controllers and stick sensitivity/deadzone profile customization for the DualSense Edge controller (note that PlayStation Accessories does not document a built-in zero-point or outer-boundary calibration wizard for standard DualSense controllers).
     • Steam Input (Windows, macOS & Linux): Provides software-level inner and outer deadzone adjustments for games launched through Steam.
2. 2
   Follow Up Resting Center Offset on the Controller Deadzone Test:

   • What It Checks: If your outer 72-sector trace looks balanced, but the stick settles away from (0.000, 0.000) when released after flicking Up, Down, Left, and Right, open the Controller Deadzone Test. You can adjust the interactive 0%–35% inner deadzone slider while comparing Raw API Magnitude against Post-Deadzone Output to identify an appropriate in-game inner deadzone setting.