Learn

How cricket analysis actually works

What a phone camera can measure about a delivery or a shot, how it is measured, and where the limits are.

Cricket analysis used to mean a biomechanics lab: reflective markers taped to a player’s joints, several synchronised cameras, and a booking. The data was genuinely good and almost nobody could get to it. Markerless analysis replaces the markers with a model that locates joints directly in the image, which means ordinary video from a phone becomes the input instead of a lab session.

Two things then decide whether the numbers are worth anything, and neither is the software. The first is where you put the camera: a lens measures movement across its sensor plane well and movement along its viewing axis poorly, so filming square to the action rather than from behind it changes results materially. The second is frame rate, which sets how finely fast movement is sampled — 60fps resolves a delivery that 30fps blurs past.

The pages below cover each measurement in turn: what it is, how it is recovered from video, and what it cannot tell you. They are written to be useful whether or not you use Stance, because a measurement you cannot interrogate is not one you should trust.

Filming checklist

The same four rules apply to every measurement on this site, and getting them right matters more than any setting inside an app.

  • Square to the action. Not behind the bowler, not behind the batter. Movement toward the lens is the axis a single camera knows least about.
  • Stumps in frame. Stumps are a regulated size, so they are the known reference that converts pixels into real distance. Without them there is nothing to calibrate against.
  • 60fps or better. At 30fps a fast delivery travels about a metre between frames, which coarsens everything derived from it.
  • Camera still, light good. A fixed surface beats a steady hand, and short exposures are what freeze a moving ball into something trackable rather than a streak.