Updated
Did the shot leave while you were still moving, did you stop and then wait, or were A and D down together? Start with Shots fired while moving, then open Counter-strafe timing in Advanced Deep-Dive, under Movement. The examples below are illustrations, not real player data.
This page is about the counter-strafe numbers a NextFrag report actually shows. For broad fundamentals see the CS2 aim training guide. For where the crosshair rested, see CS2 crosshair placement.
Tapping the opposite strafe key stops you sooner than letting go and coasting. The report reads your speed at the shot, and the A and D timing it can see before the shot.
Shots fired while moving is the share of rifle shots where your speed had just been above 88 units/s and you were still above that speed when you fired. Crouched rifle shots are left out. Lower is better.
Too early (moving) uses a lower 50 units/s line and includes SMGs, so the two numbers will not match. Lower is better.
Too late (>300ms) is the share of rifle and SMG shots where you switched strafe keys, or let the strafe key go, before the shot, and then waited longer than 300 ms after both strafe keys were up. Lower is better.
Ø overlap (too early) is how long A and D were held together, averaged only over the shots where they overlapped. The row says not measured when no shot overlapped. Lower is better.
Ø gap at shot and Typical gap (middle half of shots) skip the waits longer than 300 ms, so a high Too late (>300ms) can sit next to a small Ø gap at shot. The report prints Target gap: 15–80ms under those rows. A gap of zero can still be On time (clean), and a short gap is not a fault the report scores.
Counter-strafe taps /sec is how often per second you press A or D before rifle shots in a fight, and a higher number is the better direction, but it does not say the stop was clean.
In game: you tap the opposite key and shoot in the same motion. The crosshair was on them, and the shot left while you were still sliding. Replay one peek and check whether the click comes before your speed has actually died.
In the report: for example, Shots fired while moving of 60% means most of those rifle shots were still above 88 units/s. Acc: clean vs moving of 40% / 15% would mean the on-time shots connected more often than the still-moving shots in this demo. Treat that pair as a comparison, not a third timing score. It can disagree with Too early (moving). On time (clean) is about the keys, so a shot can be on time and still be moving. The rifle card is Counter-strafe: stop fully before you shoot (rifle). The other card is Shooting while moving.
Ten-minute drill: spend two minutes in the aim_botz workshop map with cl_showpos 1 on, so you can see vel. For six minutes, strafe one way, tap the opposite key, and shoot only when vel shows 0. Split the reps across both sides. Spend two minutes at a normal peek pace, still waiting for 0. It is the wrong drill if your problem is waiting after the stop.
Recheck with the next demo: look for a lower Shots fired while moving and a lower Too early (moving). One demo is a review, not proof the drill worked.
In game: you counter-strafe, feel planted, then take an extra look before you click. The duel is often already decided. Check a few kills you lost after you were already stopped.
In the report: Too late (>300ms) is the row to trust. For example, 30% means about three in ten of those shots waited longer than 300 ms. Do not clear yourself with Ø gap at shot or Typical gap (middle half of shots). Both skip every gap longer than 300 ms, so they can look ordinary while Too late is high. The card is Counter-strafe: fire the instant you stop, don't hesitate.
Ten-minute drill: same map, same cl_showpos 1. Spend two minutes finding the moment vel hits 0. For six minutes, strafe, tap the opposite key, and shoot that moment. Do not pause to double-check the number. Spend two minutes making the stop and the shot one motion. This drill is the opposite habit from the first one. There you wait for the stop. Here you refuse the extra beat after it.
Recheck with the next demo: Too late (>300ms) should be lower on a similar set of counter-strafe shots. Read the pattern across several demos.
In game: A and D are down together for a beat, so the stop feels mushy, or you release and then wait a moment that is not quite a full pause. Confirm it on a quiet peek before you blame aim.
In the report: for example, Ø overlap (too early) of 40 ms means the shots where both keys were down held them together for about 40 ms. not measured means it did not happen. The card is A+D key overlap. A Typical gap (middle half of shots) well above Target gap: 15–80ms, with Too late (>300ms) still low, means the waits are long but still inside 300 ms.
Ten-minute drill: on an empty practice server or in aim_botz, spend two minutes going slowly. Press A, release fully, then press D, and release fully. There should be a brief moment with neither key down. For six minutes, take a shot just after that release, without both keys down and without a pause. Spend two minutes at pace.
Recheck with the next demo: Ø overlap (too early) should be lower, or still not measured. Typical gap (middle half of shots) should sit closer to Target gap: 15–80ms from above, without a new pile of Too late (>300ms) shots. Read both rows. A single demo can move because the fights changed.
Use the labels the report displays, and keep their samples apart:
For a wider view of what a demo can and cannot prove, see CS2 demo analysis limitations. To sequence one habit across a week, see demo to training plan.
Check your counter-strafe from a demo