The DigiCue is continuously acquiring accelerometer and gyroscope samples at a sample rate of 104Hz. When the algorithm detects a ball impact event, it sets a timer. When the timer expires, it freezes the acquisition and looks back at the snapshot of the cue motion before and after impact. The furthest back in time observed is for analysis of the backstroke pause. Data from practice strokes before this have already been over-written.Straight, as in, minimal vertical deviation from a straight line?
Obviously one would not expect that result based on the practice strokes but I will not dispute objective data.
Straightness/Steering is measured in a range of 5-50ms before impact and looks for radial forces applied during this time... very quick.
Jab looks for a negative axial force during this 5-50ms window.
Follow Through measures how the cue re-accelerates, and not the actual distance the cue pushes through the bridge hand. Through the two are related.
And etc.
Since the device uses a MEMs inertial momentum chip coupled directly to the cue, it has a sensitivity beyond the effective photodiode pixel width of a common video camera.
The app takes all of this information and tries to simplify it by making a bar chart, with "good" and "bad" for the layman, based on the selected settings.