Matched estimator
Equal experimental conditions let the unknown base rate cancel in the log ratio. Interleave A/B runs and record temperature, vibration, and timing drift.
MATCHED-RUN EVIDENCE WORKBENCH
Did the clock change, or did the coin?
Normalize two experimental runs, estimate the identifiable rate signal, and keep a faster event clock separate from a shifted outcome probability. The result is an audit aid, not proof of a new field.
Pair the runs
Predict before computingRun A REFERENCE
Run B MODULATED
Computed diagnosis
AWAITING MATCHED RUNS
Choose a predictionRATE A-
RATE B-
RATE RATIO-
eta * DeltaE-
OUTCOME BIAS DELTA-
Delta Lambda-
Why it matters: total events per exposure measure the clock; positive outcomes per total events measure the coin.
Prediction feedback appears after analysis.
Evidence limit: matched arithmetic cannot exclude unmeasured drift, identify eta and E separately, or substitute for blinded primary data and independent reproduction.
Worked contrast
The bundled sample changes total events per second while both outcome shares remain 0.500. Raising only Run B positives to 126 exposes a separate 0.200 bias shift.
Claim boundary
A nonzero estimate is a candidate pattern. It does not prove MQGT-SCF, the E field, the hidden S source, or external experimental survival.