Stick Circularity Error: Why Diagonals Feel Wrong

Push your left stick straight up. Your character runs at full speed. Now push it up and to the right, into the corner of the gate, as hard as the plastic allows. On a lot of controllers, your character now runs slightly slower than they did going straight up.

That should not happen. Full deflection is full deflection regardless of direction. The gap between what you expect and what the controller reports is circularity error, and it is one of the few controller faults that shows up as a gameplay problem long before it shows up as an obvious defect.

What circularity error actually is

Your stick reports two numbers, X and Y. Together they describe a point. When you sweep the stick around the outer edge of its gate, that point traces a shape.

On a perfect controller the shape is a circle. Every direction reaches the same distance from centre, so full deflection means the same thing whether you push north, east or northeast.

On a real controller it is not a circle. It might be pinched at the diagonals, so the corners fall short. It might bulge into a square, so the diagonals overshoot. It might be an oval, longer in one axis than the other. Circularity error is how far that traced shape deviates from a true circle, usually expressed as a percentage.

The maths behind the two failure directions is worth a moment because it explains why both exist.

If X and Y are each calibrated independently so that each maxes out at 1.0, then pushing fully into a corner gives you a point at (1.0, 1.0). Its distance from centre is the square root of 2, about 1.41. That is 141 percent of full deflection, which is impossible in any sensible interpretation, so games clamp it back down to 1.0. That is the square gate case, and it produces overshoot.

If instead the physical gate is a circular ring that stops the stick at the same radius in every direction, but the sensors do not quite reach their calibrated maximum on a diagonal, the corner reads something like 0.88. That is undershoot, and it is the more common problem.

Why it makes diagonals feel wrong

Undershoot is the one that produces the complaint in the title, and it does so in a few different ways.

Threshold effects. Many games use stick magnitude as a switch. Push past 90 percent and you sprint. Below that you jog. If your diagonals top out at 88 percent, you cannot sprint diagonally. Not slowly, not sometimes. Not at all. The game is behaving correctly and your controller is lying to it.

Turn rate. In games where camera speed scales with stick magnitude, your maximum turn speed varies with direction. Sweep the stick in a circle and the camera traces an oval. Most people register this as the sensitivity feeling inconsistent rather than as a hardware issue.

Movement speed. Where speed scales with magnitude rather than being a threshold, you simply move slower diagonally, which in most 3D games is the direction you spend most of your time moving.

Overshoot causes the opposite set of problems. Because the game clamps anything past 100 percent, a whole region of physical stick positions near the diagonal all report the same maximum value. You get a dead area at the corners where moving the stick does nothing, and the transition into it feels like the stick snapping to the diagonal.

Rocket League is where people notice this most acutely, because air roll and directional control both depend on precise diagonal magnitude and the game gives you no aim assist to hide it. Our Rocket League deadzone guide covers the settings side of that specifically.

The other thing people mean by "diagonals feel wrong"

Worth separating, because two completely different problems produce the same complaint and the fixes are unrelated.

Circularity error is about the outer edge of the stick's range. It affects what happens at or near full deflection.

Deadzone shape is about the centre. If your game applies an axial deadzone, meaning it zeroes X and Y independently below a threshold, then near centre a diagonal push has to overcome the deadzone on both axes at once. The result is that small diagonal movements register later than small cardinal movements, and the stick feels like it snaps to the four cardinal directions before it will go diagonally. A radial deadzone, which considers the combined magnitude instead, does not do this.

If your diagonals feel bad during small precise movements, look at deadzone shape. If they feel bad at full push, look at circularity. Our deadzone tuning guide covers the shapes and which games expose the choice.

How to measure yours

Two minutes, no equipment.

Open a deadzone visualizer, which plots your stick position as a point on a two dimensional field and is the right tool for this because circularity is fundamentally a shape problem rather than a numbers problem.

Then:

1. Push the stick fully into the gate and hold it there
2. Keeping constant outward pressure, slowly rotate it all the way around, three full revolutions
3. Watch the shape the trace makes

What you are looking at:

A clean circle. Your stick is well calibrated with a good gate. Nothing to do.

Pinched at the diagonals, like a four-petal flower. Undershoot. The corners are not reaching full magnitude.

Square or rounded-square corners. Overshoot. Diagonals exceed 100 percent and are being clamped.

An oval. One axis reaches further than the other. This is a calibration problem rather than a gate problem and is the most fixable of the three.

Lopsided or off-centre. Your centre point is wrong, which is a separate issue and usually the first thing to fix, since it makes every other measurement unreliable.

For a number rather than an impression, note the smallest and largest distance from centre that the trace reaches. The difference between them, as a proportion of the largest, is your circularity error. Under 5 percent is good. Around 10 percent is typical and mostly unnoticeable. Above 15 percent and you are losing real capability at the corners.

Where it comes from

Gate geometry. The plastic ring that physically limits stick travel. Some are round, some are subtly octagonal, and some are round but with a larger radius at the cardinals than the diagonals because of how the housing is moulded.

Independent axis calibration. If X and Y are each normalised to their own maximum without reference to each other, the diagonal is whatever it happens to be. This is why calibration can fix an oval but not a pinched circle.

Gimbal geometry. Two-axis gimbals move each axis around a different pivot. The geometry is not perfectly symmetric, and at large deflections the error grows.

Sensor non-linearity at extremes. Potentiometers are least accurate at the ends of their travel, which is precisely where circularity is measured.

Wear. As tracks wear, an axis can lose a little range. That shows up first as an oval and later as generally reduced reach.

Replacement modules. This one catches people. Aftermarket stick modules, including magnetic replacements, sometimes have a different gate radius or different electrical range from the originals. Installing them can introduce circularity error into a controller that did not previously have it, which is a frustrating outcome for what was meant to be an upgrade. Our analog stick optimisation guide covers what to check after a module swap.

What you can actually do about it

Recalibrate first. This will not change the shape of your gate, but it will fix an oval and correct an off-centre origin, and both of those make circularity look worse than it is. Start here because it costs nothing and rules out the easy explanation. The joystick calibration tool handles it for most devices.

Use software correction for undershoot. Several remapping tools expose a control that converts a circular stick output toward a square one, usually called something like square stick or circularity adjustment. Applied at a modest strength, it scales the diagonals up and can recover the last ten percent you are missing. Applied heavily it creates overshoot, so increase it gradually and re-test rather than setting it to maximum.

Lower in-game thresholds. If a game lets you set the sprint or full-turn threshold, dropping it below your worst-case diagonal magnitude sidesteps the problem entirely. Less elegant than fixing the input, and it works immediately.

Match the deadzone shape to the game. If the game offers radial and axial deadzone options, radial is generally the better choice and removes the centre-diagonal problem described earlier.

Clean the gate. Dust and debris around the restrictor ring can prevent the stick reaching its full travel in specific directions. If your trace is pinched in one place rather than symmetrically at all four diagonals, look at the hardware before the software.

Replace the module. For a genuinely bad gate, this is the only real fix. Check what module the replacement uses and whether other people have measured circularity on it, rather than assuming any replacement is an upgrade.

Why no manufacturer publishes this

Same reason nobody publishes jitter figures. There is no standard test, there would be no way to compare numbers between brands, and a good result requires a paragraph of explanation while a bad result is invisible to anyone not looking for it.

The practical consequence is that two controllers with identical spec sheets can behave measurably differently at the corners of the gate, and the only way to know which one you have is to trace it yourself. It takes two minutes and it explains a category of complaint that most people never manage to put into words.