Ask a fighting-game player why online feels wrong and you usually get a feeling rather than a figure. “My punishes don’t come out.” “It feels heavy.” “The game decided I blocked late.” Underneath those complaints is a stack of delays the player never sees: the display, the controller, and the netcode. Each is measured in milliseconds, each is a different piece of hardware or software, and they add up.
For disabled players the stack matters twice over. Latency makes timing unpredictable, and a player whose hands or reactions are already slower has the least room to absorb that unpredictability. The good news is that the two halves of the problem — reducing delay and making inputs forgiving — are now the same conversation.
What this covers: How display lag, controller latency and netcode delay combine, how each is actually measured, and which accessibility settings make inputs forgiving enough to survive the delay. Why it matters: You cannot tune a term you cannot see. Most “online feels bad” complaints are a local hardware problem wearing a netcode disguise. Who should read this: Players building a low-latency setup, and anyone who needs a reduced-input mode to play at all.
Three delays, one button press
Total felt latency from press to pixels is a sum, not a single number. Our frame-meter guide states the stack plainly: controller poll, plus engine input lag, plus display lag, plus netcode delay. Only the last term is the network’s. The first three are local and behave identically offline — which is why a fighting game can feel crisp in training mode and wrong in ranked on the same console.
| Layer | What it measures | Typical contribution | How it is measured |
|---|---|---|---|
| Display | Signal received to pixels lit | Game-mode TVs mostly under 20 ms | Leo Bodnar lag tester |
| Controller | Button or stick movement to report | About 1.5 ms up into the teens | Gamepadla latency database |
| Engine | Input report to simulation read | Platform- and patch-dependent | In-game frame meters |
| Netcode | Remote input arrival to simulated | 0 to several frames, scaled by ping and loss | Match indicators, patch notes |
A second reference point matters for scale: fighting games run their simulation at 60 frames per second, so one frame is 16.67 ms. That means a single 20 ms display term is already more than one frame of the entire game.
How display lag gets its number
The number on a TV spec sheet is a measurement, not a property, and the measurement method decides the figure.
The standard instrument is the Leo Bodnar input lag tester. It sends a test pattern over HDMI and reads, with a photo sensor, how long the display takes to show it. Leo Bodnar’s own documentation notes it “measures combined input lag and pixel response time”, outputs a 60 Hz signal, and recommends reading the middle of the three on-screen reference bars. Because LCD panels update top to bottom, the bottom bar can never read below one frame — roughly 16 ms — at 60 Hz even on a display with no processing delay at all.
That is the trap in reading review numbers. Because the tester includes frame transmission time, a “0 ms” display is impossible to report, and figures across sites are only comparable if they use the same bar and the same settings. TechRadar explains its own process this way: it puts the TV into game mode, connects to an HDMI 2.1 port where available, and holds the sensor to the middle bar to approximate an average. In its own results, it reports most TVs under 20 ms, with LG sets averaging around 12 ms and dropping to roughly 9 ms with the game “Boost” setting, while citing a Sony X90L at 18.9 ms and a Samsung QN85C at 9.8 ms. It also notes Intel’s guidance that professional gamers look for input lag below 15 ms.
DisplayLag, which maintains a public input lag database, grades displays on a traffic-light scale: 0–20 ms is “Excellent”, 21–41 ms “Great”, 42–62 ms “Okay”, and 63 ms or higher “Bad”. Its own advice for 60 Hz play is blunt: “Stick to 10 ms or lower if you want the best performance at 60hz.”
The practical upshot for a fighting game is narrow. Game mode and disabled motion smoothing are the two settings that matter most; both are free. After that, the difference between a 9 ms and a 19 ms panel is about half a frame — real, but smaller than the errors a bad network introduces.
The controller is a variable, not a constant
Controller latency is where the widest spread lives, and where accessibility hardware and latency hardware are the same purchase.
Gamepadla maintains a community-tested database of button and stick latency across hundreds of pads, sticks and adapters, and it separates wired, dongle and Bluetooth figures. Its results are not uniform: the database lists a 1.54 ms button-latency figure for the Mad Catz Street Fighter V Arcade FightStick Tournament Edition 2+ and 8.31 ms for the Hori Fighting Stick Alpha XS. Those are different devices in the same genre, and the gap between them is comparable to the spread across an entire generation of televisions.
Two caveats keep those numbers honest. First, Gamepadla’s own disclaimer states it measures the delay between successive changes in stick position rather than the traditional press-to-response latency, and that results should be “used as a guide, not as an exact measurement of input latency.” Second, the connection matters as much as the device: the database files separate wired, wireless-dongle and Bluetooth timings, and Bluetooth is routinely the slowest of the three.
For a fighting-game player the controller decision carries an extra weight, because it is also the accessibility decision. Remapping, one-handed layouts, leverless sticks and macro buttons all live at this layer. That is the subject of the second half of this article.
Netcode is the term you cannot fully control
The last term in the sum is the network’s, and it behaves differently from the others. Rollback netcode predicts the opponent’s inputs and re-simulates when the prediction is wrong, so it hides consistent latency well and packet loss badly. That trade is covered in depth in our rollback explainer; the point here is that the netcode term is real delay you feel, and the developers keep tuning it.
Bandai Namco’s Tekken 8 patch 1.04, for example, listed “Optimization of packet transmission and reception as well as rollback processing during online matches” among its functional improvements, alongside an improved indicator for detecting processing delay. Tekken 8 also exposes a rollback preference setting — players can select “Prioritize Response” in the game options — and its official PC troubleshooting page walks players through the in-match red indicator that tells you whether your machine or the opponent’s is the one causing a processing failure. That indicator is the single most useful diagnostic in the stack: it separates “my local setup is fine and the network is bad” from “my PC is dropping frames and generating the lag myself.”
The official guidance also shows how local and network problems blur. The troubleshooting page asks players to check display, system specs, in-game graphics settings and PC settings in turn, because a machine that cannot hold 60 frames per second will produce rollback corrections that look exactly like a bad connection.
What disabled players actually need
If latency is the problem, accessibility features are the buffer. Three design choices in the current fighting-game generation matter most, and each has a different relationship to latency.
Reduced-input modes remove the execution demand. Capcom ships three control types in Street Fighter 6: Classic, Modern and Dynamic. The official description calls Modern and Dynamic “brand-new options designed to improve accessibility for first-time players as well as people who haven’t played a Street Fighter game in a long time.” Modern maps special moves to a single button plus a direction. Dynamic goes further and performs attacks and combos automatically from a single Auto-Attack button, but it is “limited to certain modes in Fighting Ground” and “cannot be used in online matches.”
There is a cost. Street Fighter 6 director Takayuki Nakayama told WIRED that under Modern controls “special moves only deal 80 percent of their overall power”, and players cannot select the light, medium or heavy version of a move. Access-Ability’s hands-on with the scheme, written by a player with dyspraxia and ADHD, judged the trade worthwhile: each character was missing roughly two or three moves, but the reduction in input complexity made deliberate, reliable play possible.
Bandai Namco’s equivalent, Tekken 8’s Special Style, is described as combining and evolving the Easy Combo and Assist features of Tekken 7, assigning each button a recommended move and combo that changes with the character’s state. Reviewers found it notable that it can be toggled in and out by a single button press at any point in a match, that characters auto-block regular attacks while standing still, and that there is “no penalty for using it across all modes.”
For a player with a motor delay, the latency benefit is indirect but real: an auto-combo is timed by the game’s buffer rather than by the player’s hands, so the sequence itself no longer has to survive network jitter frame by frame.
Remapping and one-handed layouts are the difference between playing and not. Can I Play That’s Street Fighter 6 review notes that controls are customizable across modes — battle settings for each player side, avatar battle type, overworld controls and other input options. In its Tekken 8 review, the same outlet found that “controls are completely customizable” and that “all button combinations can also be programmed to a single press,” which is exactly what a one-handed layout needs. It also flagged an oversight that undercuts the feature: the controller customization menu is the one screen in Tekken 8 that still requires the D-pad to navigate, so a player who needs remapping to avoid the D-pad must use it to get to the remapping screen.
PC Gamer’s account of the same design shows what the feature buys: the writer, who has Klumpke’s palsy and a delayed, weakened right arm, described binding the auto-combo to the right bumper and later moving grab off a trigger and adding a macro for a quick-step motion — adjustments that turned a six-button execution problem into a scheme one hand could run.
Audio and haptic cues replace reaction tasks that latency makes worse. A cue that requires a fast visual reaction is the most latency-sensitive input in the game; a cue that carries information regardless of timing is not. Capcom worked with the barrier-free esports organisation ePARA on Street Fighter 6’s sound accessibility, and ePARA reports that three totally blind players tested the result. The features include a continuous sound that changes pitch and interval to signal distance to the opponent, plus distinct cues for high/mid/low attacks, cross-ups, and gauge levels.
Reviewers were clear about the ceiling. Can I Play That’s Street Fighter 6 review praised the “comprehensive audio cues” and fully customizable HUD but listed “no menu narration” as a major con, because a blind player needs sighted help to reach the menus where the cues are enabled. Its Tekken 8 review found a dedicated accessibility menu with only one option inside it — a color perception filter with intensity slider — and audio settings with no subtitle customization, which is a problem when characters speak different languages.
| Need | Street Fighter 6 | Tekken 8 |
|---|---|---|
| Reduced-input mode | Modern (one-button specials); Dynamic (auto-attack, offline only) | Special Style (button-assigned moves and combos) |
| Cost of the mode | Modern specials at ~80% power; fewer moves | No penalty; togglable any time |
| Remapping | Customizable per mode and per side | Fully customizable; combos to single press, but D-pad-only customization menu |
| Non-visual cues | Distance, attack height, cross-up and gauge sounds (built with ePARA) | Distinct sounds per character; audio cue on Rage and Heat activation |
| Menu access | No menu narration | One accessibility menu option; no subtitle customization |
The compound problem
Put the layers back together and the accessibility case becomes a latency case. A player with a slow display, a Bluetooth pad and a lossy connection is not adding three independent annoyances; they are adding three terms that all shorten the same timing windows. Reduce any one term and every remaining reaction window improves by that much.
The reverse is also true, and it is the part that gets missed: input simplification and latency reduction are substitutes for the same goal. A game that removes the need to hit a frame-perfect sequence is doing for a disabled player what a low-latency setup does for everyone — widening the window in which an action still counts. That is why the strongest accessibility work in fighting games is not a settings menu off to the side; it is the same engineering that determines whether the game feels good online.
A local-latency checklist
- Put the display in Game Mode and disable motion smoothing and interpolation.
- Use a wired connection for the controller, or check the wired figure before trusting a wireless claim.
- Use Ethernet, not Wi-Fi, for online play; jitter feeds the netcode term far more than ping does.
- Hold a locked 60 frames per second; a frame-rate drop becomes rollback.
- Watch the in-match connection and processing indicators to tell your own hardware lag from the network’s.
- If you need one, enable the reduced-input mode and remap to a layout your hands can actually run — then accept that the mode and the low-latency setup are doing the same job.
Next read
- For how the delay is auto-tuned rather than set by hand, see Street Fighter 6 Frame Meter and Input Delay Tuning.
- For the prediction-and-correction loop underneath the netcode term, see Rollback Netcode in 2026: Prediction and Its Limits.
- For why returning to a game every season gets expensive, see Season Cadence Math: 6, 8 and 12 Weeks to Burnout.
- For picking a character whose inputs fit your hands, see Street Fighter 6: How to Pick a Main by Archetype.