Sub-tick Isn't Magic: The Physical Math of Counter-Strafe Windows, Crosshair Recovery, and CS2 Movement

November 16, 2026 at 14:00:09

A technical breakdown of CS2's sub-tick movement model: why counter-strafing is a timed window, how input sampling and velocity integration diverge, and why you should calibrate with bullet holes instead of your crosshair.

Introduction

Most players treat a counter-strafe as a single key action: release W, tap S, shoot. But in CS2's movement model, a counter-strafe is a time window with a clear start and end, and sub-tick sampling decides which frame the edges of that window actually land on. Press too early and the reverse velocity hasn't been eaten yet; press too late and the crosshair has recovered while your character is still sliding. This article is about the math, not feel-based mysticism.

1. Sub-tick sampling slices a single frame into many pieces

First, clear up a common confusion: CS2 servers still advance world state at a fixed tick rate, but movement and input handling introduce sub-tick sampling granularity. For players, there is only one key conclusion — your input is no longer quantized to whole frames, but velocity integration still steps with the world.

This means the physics of a counter-strafe can be split into two stages:

  1. Input stage: the moment you press the reverse key, the input may be sampled earlier or later than the world step.
  2. Integration stage: acceleration decays horizontal velocity over several sub-steps until the speed magnitude drops below the precision threshold.

The first stage decides when you start braking; the second decides how long it takes to stop. The boundaries of the two stages don't coincide, which is why the same counter-strafe feels noticeably different under different network conditions — you control the input stage, the server settles the integration stage.

Worth noting: velocity decay is not linear. The higher the horizontal speed, the larger the absolute value eaten per unit time; as speed approaches the threshold, decay actually slows down. This explains a common phenomenon: the last short slide of a counter-strafe is the most annoying, because that is exactly the interval where decay is slowest, and crosshair precision checks land right inside it.

2. The real structure of a counter-strafe window: three stages, not two

Breaking a counter-strafe into "press reverse — stop — shoot" is oversimplified. In practice it has at least three stages:

  • Reaction stage: from deciding to stop to the finger actually pressing, usually 60–120ms, varies by person.
  • Braking stage: from reverse input taking effect until horizontal velocity falls into the precision threshold, depends on current speed and direction angle.
  • Recovery stage: after speed qualifies, the crosshair returns from spread state to acceptable shooting precision.

Most people only train the braking stage, and end up losing in real matches at the junction of reaction and recovery. The reaction stage can't be compressed much, but the recovery stage can be anticipated: if you know roughly how many sub-steps the braking stage needs, you can enter shooting preparation during the latter half of braking instead of waiting until it "feels stopped" to move the mouse.

Here's a counterintuitive point: diagonal counter-strafes are harder than straight ones, not because they need more keys, but because after decomposing diagonal velocity, the two axes decay asynchronously. One axis qualifies first while the other is still sliding; crosshair checks look at composite speed, so you flip back and forth between "looks stopped" and "actually not stopped".

3. Crosshair recovery is a check, not an animation

In CS2, the visual crosshair recovery and the shooting precision check are two separate things. Visual recovery is faster, the check is slower, and the gap between them is the source of many "the crosshair was back but the shot flew off" moments.

In practice, build your own calibration line: in an offline map, do fixed-distance counter-strafe shots at a wall and record the turning point where bullet impacts go from spread to clustered, instead of staring at the crosshair. Calibrate with bullet holes, not the crosshair. The crosshair is your reference frame; bullet holes are the server's answer.

One more training suggestion: train counter-strafing as a separate subject, don't mix it with spray control. First train "first shot always hits after stopping", then "three-round burst after stopping", and only then moving-to-contact. Reverse the order and you'll drag braking-stage error into your spray rhythm, getting messier the more you practice.

4. Settle the math first, then talk about feel

Sub-tick didn't make counter-strafing simpler; it made the sources of error more separable. Reaction stage relies on anticipation, braking stage on muscle memory, recovery stage on bullet-hole calibration. Each stage has its own training method; mixing them only yields a vague "feel is okay".

Next time you browse skins on C5GAME or scroll through wear values on the Steam market, think about this same math: what you see is the pattern, what you can calculate is the timing. Aiming is the same.