· Failures · Claim-safe

Fixed false-alarm residual detection: why GSRF loses

Restates locked Evidence / workbench numbers only. Not universal performance.

Teams sometimes try to use filter residual (raw − filtered) as a cheap anomaly detector. On locked synthetic excursion packs with matched false-alarm rate:

MethodResidual recall (matched FA)
GSRF Practical~0.44
EMA~0.83

Why peak compression hurts detection

GSRF pulls peaks toward normal and leaves a different residual structure in quiet regimes. Matching FA forces a higher threshold; recall falls. That is not “EMA is smarter”—it is a different objective.

Do this instead

Verdict

Red light for fixed-FA residual alarms. Green light for oscillation/peak calm. Cage: Evidence boundaries · When GSRF lost.

Sources: probes v5 alarm · Evidence safety cage · claim ladder

Questions people actually ask

Can GSRF residual be a fixed-FA detector?

Not as a product claim. On locked probes GSRF residual recall was ~0.44 vs EMA residual ~0.83 at matched false-alarm rate.

What should I use for residual alarms?

Keep a dedicated residual/detector design (often EMA residual or specialist FA tools). Do not force GSRF to be the alarm engine.

Does peak compression help detection?

Different job. Peak compression toward normal is not fixed-FA event detection performance.

Related research