Skip to content

Reading Results

The results screen appears when a session completes, under the heading Session complete.

If the session converged, a green card shows:

  • Anticipation threshold — e.g. 133.5 ms. The headline number
  • Standard error — e.g. ± 12.4 ms. The precision of that estimate

If it did not converge, an amber card shows No threshold — session did not converge, with one of two reasons:

ReasonMeaningWhat to do
Fewer than 6 reversalsThe staircase never settledRun 60 trials rather than 20 or 40
Too scattered to support a thresholdPerformance varied during the sessionCheck for fatigue, distraction, or strategy change. Rest and rerun

No number is emitted in this case, not even a caveated one. See How the Staircase Works.

Always shown, converged or not:

RowUnitHow to read it
TrialscountHow many ran
Accuracy%Raw proportion correct. Not comparable across sports
Accuracy (guessing-corrected)%Chance removed. Use this one for any comparison
No-response rate%Timeouts. The commitment measure — report separately, always
Median RTmsTime from blackout to touch
ReversalscountStaircase direction changes. Needs ≥ 6; more means a more reliable estimate
Timing-invalid trialscountAmber, shown only if > 0. Trials whose blackout missed by more than two display refreshes. Automatically excluded from the threshold

A handful of timing-invalid trials is normal. Many of them means something is wrong — the app was backgrounded, the device thermally throttled, or the instrument needs re-verifying. See Timing Verification.

Larger is better. A higher threshold means the athlete needs less information, earlier.

A change is only real if it exceeds roughly 2 × the standard error. A move from 133 ± 12 ms to 141 ± 14 ms is noise, not improvement.

Do not compare across sports. Different chance levels, different tasks.

Do not compare across viewpoints. Side-on footage measures a different skill.

Do not compare across footage sets. A threshold measured against one opponent’s technique is not directly comparable to another’s — you have changed the stimulus, not just the athlete.

Do compare the same athlete over time, against the same footage set. That is what the standard error is for.

These two must be read as a pair:

PatternInterpretation
Good accuracy, low no-responseReading well and committing. The target state
Good accuracy, high no-responseA commitment problem, not a perception problem. They can read it but won’t commit under time pressure. Different intervention entirely
Poor accuracy, low no-responseCommitting on insufficient information. Perception is the limiter
Poor accuracy, high no-responseTask is too hard, or the athlete has disengaged. Check the session converged at all

Tap Saved to return to the setup screen. The session was already written to the device database — the button is a confirmation, not an action.