A rolling ten-round window, explicit LOW/HIGH indicator behaviour and the 1-4-4-1 structure — evaluated the same way every single round, with an evidence gate that blocks any projection the data does not support.
Every round runs the same 20-step derivation: classification, start detection, direction, indicator behaviour, projection. Same input, same output, always.
Mandatory evidence gate
A projection is only shown when the start point, direction and indicator behaviour are all proven from the window. Otherwise the engine reports UNCONFIRMED.
Live or replayed
Record rounds by card face, result only, or from a scoreboard screenshot — then replay any shoe round by round to audit what the engine knew at each step.
Rules you control
Everything the methodology leaves undefined is an explicit parameter. Left empty, it blocks decisions rather than guessing on your behalf.
How a round is processed
01Record the hand — card faces, result only, or a screenshot import.
02Classify the losing side as a LOW or HIGH indicator.
03Locate the start point and score NORMAL against REVERSE.
04Derive CUT or FOLLOW behaviour for each indicator separately.
05Run the evidence gate over every requirement.
06Publish a projection, or state plainly that it is UNCONFIRMED.
This tool is an analysis and record-keeping instrument. It reports what the recorded data supports and nothing more; it does not predict outcomes and makes no claim about future results.