Eight-Direction Accuracy Test
Most games do not read your analog stick as a smooth circle. They chop it into eight slices and ask which one you landed in. Menus do it. Weapon wheels do it. Fighting game motions do it. So the question that actually decides whether an input works is not how precise your stick is, it is whether it puts you in the slice you were aiming for and keeps you there. This test names a direction, waits for you to push, and then measures the angle you delivered, how far it sat from the boundary where the game would have read something else, and whether it stayed put while you held it.
Run the Accuracy Drill
Setup
Target
The thin grey lines are where one slice ends and the next begins. Land near the middle of a wedge and the game reads you clearly. Land near a line and a small wobble sends the input somewhere else.
Results
Scorecard
Every direction gets its own row, because sticks almost never fail evenly.
Report
How to Run It Properly
- Use a cable if you have one, and press any button on the pad so the browser can see it.
- Hold the controller the way you actually play. Grip changes thumb angle, and thumb angle is half of what this test measures.
- Press Start. The drill names a direction and waits.
- Push toward it in one motion. Do not creep out slowly hunting for the perfect angle, because that is not how you play and it is not what the test is for.
- Hold until the bar fills. That is when the stability part gets measured.
- Return to center between prompts. The drill waits for you, so there is no rush.
- Do at least two rounds. One round tells you what happened once. Two rounds start telling you what happens reliably.
If the stick will not reach a direction at all, that is a range problem rather than an accuracy problem, and the circularity test is the page that measures it. Before either, the stick drift test tells you whether your center has moved, since every angle here is measured from center.
Why Eight Directions Matter More Than You Think
Analog sticks feel continuous. You push, the character moves at an angle, and it all seems smooth. Underneath, a surprising amount of what you do gets rounded off to one of eight compass points before the game does anything with it.
Menu navigation is eight way, or often just four. Weapon wheels and emote wheels are wedge based. Fighting game motions are read as sequences of directional inputs on a numbered pad, which is why a quarter circle is described as down, then down forward, then forward. Platform fighters check whether you crossed into a diagonal region. Even plenty of action games with fully analog movement still use eight way logic for things like ledge grabs, cover snapping and lock on switching.
What follows from that is worth sitting with. The precision of your stick barely matters for these inputs. What matters is which wedge you land in. A stick that delivers 41 degrees instead of a perfect 45 has an error of four degrees, which sounds bad until you realise the wedge runs from 22.5 to 67.5 and you are comfortably inside it. Meanwhile a stick that delivers 24 degrees has an error of 21 degrees and is sitting one and a half degrees from the boundary. The first is fine forever. The second will betray you at the worst possible moment, and it will look like your mistake.
That is what margin means
Every wedge is 45 degrees wide, so the middle of a wedge is 22.5 degrees from either boundary. Margin is how much of that cushion you have left. One hundred percent means you landed dead center. Zero means you landed exactly on a line, where the tiniest wobble flips your input to the neighbouring direction.
Margin is the number to watch on this page. Average error tells you how accurate you are in general. Margin tells you how close you are to being wrong.
What This Test Does That a Circularity Sweep Cannot
The circularity test is passive. You roll the stick around the rim and it records the shape of the gate. It is a hardware measurement and your thumb barely enters into it.
This test is active. It asks you to deliver a specific input and measures what actually arrived. That makes it a combined measurement of you and the controller together, which is a feature rather than a flaw, because you and the controller are what the game receives. Nobody plays through a circularity sweep.
The two answer different questions. Circularity asks whether the stick can reach every direction. This asks whether you can reliably put it there. A stick can have a flawless round gate and still be awkward to aim with because the axes are scaled unevenly. A stick can have a dented gate and still deliver every direction cleanly, because the dent sits where the reading has already registered.
Run both when a stick feels off. Agreement between them is how you narrow down what is actually wrong, and the section further down on separating your thumb from your hardware explains how to read the pair together.
Reading Your Numbers
Average error
How many degrees you missed by, averaged across every prompt. Under four degrees is tight. Under eight is comfortable. Past fourteen you are landing near boundaries often enough that inputs will occasionally read as the wrong direction.
Worst direction
The compass point where you missed by the most. This is often more useful than the average, because sticks and thumbs both fail unevenly. One bad direction in an otherwise clean set is a much more specific problem than eight mediocre ones.
Average margin
How much cushion you had, on average, between where you landed and the nearest boundary. Above 70 percent is comfortable. Below 40 percent means you are living close to the edge and small movements will change what the game reads.
Misreads
The number of moments during your holds where the stick had wandered into a different wedge from the one you were aiming at. This is the most direct answer the test can give you, because a misread is not a statistic. It is an input the game would have read wrong.
Wander
How much the angle moved around while you were holding it steady. High wander with a low error means you can find the direction but not keep it, which usually points at noise or an unstable grip rather than a range problem. If the wander looks suspicious, the stick jitter test measures the same instability without your thumb in the loop.
The Axis Imbalance Signature
Here is the pattern this test is best at catching, and the reason it earns its place next to the other stick pages.
Imagine your X axis reaches a full 1.00 but your Y axis has worn down and only reaches 0.85. Push toward the upper right corner with equal physical effort on both axes and you send 1.00 across and 0.85 up. That is not 45 degrees. That is about 40. Every diagonal gets pulled toward the horizontal, by the same amount, in the same way.
The signature is unmistakable once you know to look for it. Up right comes in below 45. Up left comes in above 135. Down left comes in below 225. Down right comes in above 315. All four lean toward the horizontal axis. Flip the weakness to the X axis and all four lean toward the vertical instead.
The tool checks for this automatically and tells you which axis is under reaching when it finds it. It matters because an imbalance like this is invisible in everyday play. Your cardinal directions stay perfect. Your diagonals are quietly wrong by five or ten degrees, forever, and no amount of practice corrects for a hardware bias. You just develop a slight lean and never know why.
Once the tool flags it, the circularity test confirms it from the other side, since a weak axis shows up there as an oval rather than a circle. Two tests agreeing on the same axis is about as clear a diagnosis as browser tools can give you.
Square Gates and the Diagonal Problem
Plenty of controllers have square gates, meaning the stick physically travels further into the corners than it does straight up or straight across. That is a housing design choice rather than a fault, and it interacts with this test in two ways worth knowing about.
First, your diagonal reach readings will come in higher than your cardinal ones. That is expected and it is not an accuracy problem.
Second, and less obviously, games that apply radial scaling to square gate input change the delivered angle slightly as part of squaring the circle. So the same physical thumb position can produce a slightly different angle in a game that scales than in this test, which reads the raw values. If your in game feel and your test results disagree, scaling is a likely reason, and the controller mapping tool will show you whether anything is reshaping your axes before they arrive.
Telling Your Thumb Apart From Your Hardware
This is the honest limitation of any active test, so here is how to work around it.
Look at consistency, not accuracy
Human error is scattered. Hardware error repeats. If your Up right lands at 39, then 40, then 38 across three rounds, that is the controller. If it lands at 39, then 51, then 44, that is your thumb. Run more rounds and read the pattern rather than the average.
Compare the two sticks
Test the left and the right on the same day. Your thumbs are roughly as good as each other. If one stick scores far worse, the difference is in the hardware, not in you.
Compare with a passive test
If this test says your diagonals lean horizontal and the circularity test says your gate is a horizontal oval, that is one fault confirmed twice. If this test flags a problem and every passive test comes back clean, the problem is grip, thumb placement or cap height.
Change one thing at a time
Take extended thumbstick caps off and repeat. Change your grip and repeat. Move the controller from your lap to a desk and repeat. Each of those changes only your side of the equation, so any difference in the result is yours.
What Counts as Normal
| Reading | Tight | Comfortable | Worth acting on |
|---|---|---|---|
| Average error | under 4 degrees | 4 to 8 degrees | over 14 degrees |
| Average margin | over 80 percent | 55 to 80 percent | under 40 percent |
| Worst direction error | under 8 degrees | 8 to 15 degrees | over 20 degrees |
| Misreads across a run | zero | one or two | three or more |
Cardinal directions almost always score better than diagonals, and that is normal for everyone. Up, down, left and right have a physical edge to push against and a natural thumb position. Diagonals have neither. Expect your four corners to carry most of your error, and only worry when one corner is much worse than the other three.
Where This Shows Up in Games
Fighting games. The clearest case. A quarter circle needs you to pass cleanly through down forward, and a dragon punch motion needs forward, then down, then down forward. If your down forward reads as down, the move does not come out and you blame execution. Players who struggle with one motion but not its mirror image should test both diagonals here before spending another month in training mode.
Platform fighters. Angled inputs decide which move you get, and the boundaries are often tighter than 45 degrees. Margin matters more here than anywhere else.
Menus and wheels. The most common everyday symptom. Selecting the item next to the one you wanted, or a menu cursor that skips diagonally, is almost always a boundary problem rather than a broken stick.
Twin stick shooters. Eight way aiming is common, and a stick that leans on the diagonals means your shots consistently miss on one side.
Racing and shooters. Least affected, because both mostly use continuous input. If those feel wrong but this test comes back clean, look at input lag and at the snapback test instead, since both cause problems that feel like inaccuracy but are not.
What to Do About a Bad Score
Try these first
- Take off extended thumbstick caps. Taller caps change the angle your thumb naturally sits at and add leverage. Test with and without.
- Recalibrate. On console or through vendor software. This fixes small centering errors that throw every angle off by the same amount.
- Capture the resting center here. The setup panel has a button for it. If your stick rests off center, every angle in the drill is measured from the wrong origin.
- Check what is in the middle. Steam Input, DS4Windows and vendor apps can apply scaling that changes delivered angles. Close them and run again.
If the pattern is consistent
- One axis under reaching. Range scaling in software can even the axes out. If the imbalance is large, the module is worn and a swap is the real fix.
- One direction much worse than the rest. Cross check against the circularity test. A short spot in the same direction confirms a worn part rather than technique.
- High wander with low error. The stick finds the direction but will not hold it. Measure the instability directly with the stick jitter test, and check the link with the connection stability test if the readings jump rather than hiss.
- Everything is bad on one stick only. That is hardware. Run the full controller tester to see whether anything else on that side is misbehaving too.
Worth knowing
If your diagonals are genuinely unreliable and you are playing something that depends on them, the D-pad is often the better tool for the job. Check yours is healthy on the button test before you switch, since a D-pad with a dead diagonal will not help.
Using This as Practice
Once your hardware checks out, this drill works as training. Shuffled order removes anticipation, so you are reacting to a prompt rather than running a memorised loop, which is much closer to how directional inputs actually happen in a match.
Two things are worth tracking over time. Average margin, because it improves faster than average error and it is the number that predicts whether inputs will register. And the gap between your best and worst direction, because most people have one corner they are quietly bad at and never notice until something measures it.
Five minutes a few times a week moves the numbers. Save the report each time and the trend does the rest.
Method and Honest Limits
This measures you and the controller together
Stated plainly because it is the thing people misread. A poor score is not proof of a faulty stick. The section above on separating the two is the part to read before you order parts.
Sample rate
Readings arrive in step with your display refresh, so a 60Hz screen gives roughly sixty per second. That is more than enough for measuring a held angle. It is less useful for measuring exactly when you arrived, so treat the timing side as approximate. The polling rate test shows how often your pad is actually reporting underneath that.
Grip changes results
Keep it consistent across rounds and across sessions. Comparing a lap session against a desk session tells you about your posture, not your controller.
Both sticks
They wear differently and your thumbs are not equally practised. Test each one and compare.
Browser and Platform Notes
- Chrome and Edge report axis values most consistently. Firefox works. Safari is the least predictable.
- Keep this tab focused for the whole drill. Background tabs get throttled and the sampling stops mid hold.
- Some third party pads only expose their axes properly in DirectInput mode, usually reached with a button combination involving Home.
- If you are unsure which axis numbers your pad uses, the controller mapping tool prints them and the gamepad viewer shows them moving live.
- A pad low on charge reports less consistently over Bluetooth, so the battery health test is worth a glance if your rounds disagree with each other.
- Everything runs locally in the page. Nothing about your controller or your scores leaves your device.
Questions People Actually Ask
- Is a bad score my fault or the controller's?
- Both are possible, which is why the drill runs multiple rounds. Consistent errors in the same direction are hardware. Errors scattered differently each round are technique. Comparing your two sticks settles it faster than anything else.
- Why are my diagonals always worse than my cardinals?
- Because that is true for almost everyone. Straight directions have a physical edge to push into and a natural thumb position. Diagonals have neither. Only worry when one corner is much worse than the other three.
- What is a good average error?
- Under eight degrees is comfortable for most play. Under four is tight. What matters more is margin, since that is the number that decides whether an input reads correctly.
- Does this test the D-pad?
- No, this is an analog stick test. The button test covers D-pad directions, and it is worth checking if you are thinking of switching inputs because your diagonals are unreliable.
- My margin is high but I still get wrong inputs in game.
- Then the problem is probably not where you land, it is what happens on the way there or on the way back. Overshoot on release is measured by the snapback test, and it produces exactly this kind of complaint.
- Should I use in order or shuffled?
- In order for diagnosing hardware, because it is easier to follow and you can concentrate on the stick. Shuffled for practice, because reacting to a prompt is closer to how you actually play.
- Can I use this to compare two controllers?
- Yes, and it is one of the better uses. Same grip, same session, same number of rounds, then compare the reports. Save both.
- Does it work on a phone or tablet?
- Yes, with a controller connected over Bluetooth or through an adapter. The target scales down and the drill works the same way.
Related Tests on Gpadtester
This page measures delivery. These measure the parts underneath it, and results that agree across two pages are worth far more than either on its own.
Sticks and triggers
Whole controller
Connection and response
Read next
How to fix controller drift covers cleaning, recalibration and module replacement, which is the repair path when one direction is consistently worse than the rest.
Other hardware
Away from the controller: keyboard test, mouse test, mic test, frame rate test and refresh rate test.