Why Does Flicking Feel So Weird in CS2?

First, what did not change

It is worth clearing this away, because a lot of theories start by assuming the mouse maths moved. It did not, and CS2's own console variables say so:

There is no acceleration or smoothing layer sitting between those and your crosshair. So a given sensitivity value turns you a fixed amount per unit of mouse movement, consistently. If your aim feels different, the translation from hand to view angle is not where to look. If you are moving between games rather than versions, the sensitivity converter handles that part properly.

Candidate one: sub-tick input timing

CS2's headline architectural change is that the server no longer evaluates the world only at discrete tick boundaries. Valve's framing is that the server knows the exact moment an action starts, a shot is fired or a grenade is thrown, and the stated consequence is that movement and shots are equally responsive regardless of tick rate.

For a flick, that matters in a specific way: a flick is a shot fired during rapid view-angle change, so when the click is considered to have happened determines where the crosshair was considered to be pointing. Under the old model that got snapped to a tick edge. Under sub-tick it is timestamped. Both are internally consistent; they are not the same, and the difference is largest exactly when the view is moving fastest.

Two things are worth knowing before you treat this as the answer. The first is that view angles themselves are part of the sub-tick pipeline — sv_subtick_movement_view_angles is a real CS2 convar and it defaults to on, and Valve has adjusted how those angles are transmitted so that they are sent with other sub-tick events rather than every frame they change.

The second is that this system is still being tuned. A September 2025 update lists "various improvements to sub-tick shooting consistency". An August 2025 update changed the bhop jump-spam timer to start at the instant the input is registered rather than at the end of the sub-tick that processed the command. These are not the release notes of a settled system, and if your sense of the game shifted at some point without you touching anything, that is a plausible reason why.

What sub-tick did not do is remove latency, interpolation or lag compensation — a point we go through in detail in the CS2 peeker's advantage, because it is the single most over-claimed thing about the engine.

Candidate two: what you actually see

This is the one we think is most underrated, because "feel" in a first-person shooter is largely visual feedback, and the visual feedback layer in CS2 was rebuilt.

• Every first-person animation was replaced. A July 2025 update moved CS2 to AnimGraph2 and replaced all first-person animations — deploy, firing, reload and inspect — noting that the switch required re-authoring existing game content.
• Then every third-person animation was. A beta build in April 2026 re-authored all third-person animations, in several cases adjusted in response to player feedback, and those changes went live later that month. That changes what an enemy model does as you swing onto it, which is what your flick is actually reading.
• Viewmodel motion is tied to view-angle velocity. This is stated directly in the patch notes: an August 2025 update made view angle velocity affect spectator target viewmodels the same way it affects local player viewmodels. In other words, how fast you are turning drives how your gun moves on screen.

And that last system has genuinely misbehaved. Valve fixed a case where viewmodel motion caused by view angle changes misbehaved at certain frame rates in early September 2025, and a few days later fixed a case where the same motion depended on the direction you were facing. Earlier, an August 2025 update fixed a bug that prevented first-person viewmodel lag entirely.

Read those together and a familiar complaint stops sounding like superstition: players reporting that fast turns looked wrong, that it varied with their frame rate, and that it depended which way they were facing, were describing bugs that were subsequently fixed. That is the strongest documented case for "flicking felt weird" being about something other than you.

One practical footnote while you are here: CS2 clamps viewmodel_fov between 60 and 68. If you carried a config over and set a value outside that band, you are not running what you think you are running.

Candidate three: raw input is not optional any more

CS2 reads your mouse as raw input and gives you no way to turn it off — there is no settings toggle and m_rawinput is not present in CS2's console variables at all. Our full write-up is at does CS2 have raw input.

For most people this changes nothing, because they already had raw input on. But if you were one of the players who deliberately ran without it in CS:GO — building your muscle memory on top of Windows pointer acceleration — then CS2 removed the thing your aim was calibrated against, permanently, with no route back. That is a real and total change in feel, and it is worth ruling in or out first because it has a definite answer.

Worth ruling out at the same time: acceleration configured on the mouse itself, in vendor software or onboard memory, is applied before the data reaches the game. Raw input cannot strip that out, and it will happily make flicks inconsistent while looking like a game problem.

Candidate four: your connection, not your aim

Before blaming the engine, rule out the network, because the symptoms overlap badly. Shots that seem to land but do not register, and enemies that appear a fraction late, both read as "my aim is off" from the inside.

CS2 does not give you the CS:GO-era rate commands to fiddle with — that whole approach is dead, as covered in do CS2 interp settings changes work. What it gives you instead is a telemetry HUD under Settings → Game → Telemetry, which can show ping, the rate of ticks missed to loss or jitter, and a jitter graph. That is measurement rather than guesswork, and it is the correct first move.

What to actually do about it

1. Turn on the telemetry HUD and check whether you are losing ticks before you touch anything else. A network problem will not be fixed by a sensitivity change.
2. Check your mouse's own software for acceleration, angle snapping or smoothing, and turn them off if you want genuinely 1:1 movement.
3. Prune your config. A migrated CS:GO autoexec is full of lines that no longer resolve, and a clamped viewmodel_fov is a good example of a silent difference — the CS2 console commands reference is the cross-check.
4. Then stop changing things. Pick a sensitivity and hold it. Adaptation is the mechanism that fixes "feels weird", and it cannot run while the target keeps moving.
5. Retrain deliberately rather than in matchmaking. Structured aim training gives you repetitions with feedback; ranked games give you neither.
6. Work on the part that is not mechanical at all. Most flicks exist because the crosshair was in the wrong place to begin with — how to aim in CS2 deals with that properly.

The honest summary

CS2 changed input timing, replaced the animation systems on both sides of the camera, tied viewmodel motion to how fast you are turning and then shipped fixes for bugs in that exact code, and made raw input compulsory. Any of those could plausibly account for a shooter feeling different in the hands.

What nobody has is a controlled measurement showing which one is responsible for your experience, or how large the effect is. Treat pages that hand you a percentage on this subject as fiction. The verifiable parts are above; the rest is adaptation, and adaptation only works if you let your settings sit still long enough for it to happen.