# SignalGuard 1.3 — Validation Copy

## Technical validation statement
SignalGuard 1.3 uses a strictly causal robust detection engine: the current observation
is evaluated against prior history only.

The release candidate was stress-tested across 20 random seeds and multiple controlled
scenarios, including clean signals, strong and moderate spikes, persistent level shifts,
high-noise signals, gradual drift, missing data, and stuck-sensor behavior.

Key synthetic benchmark results:
- Strong spikes: 100% recall, ~93.4% mean precision
- Moderate spikes: 87% mean recall, ~91.8% mean precision
- Persistent shift events: 100% event recall
- High-noise spikes: 100% recall, ~94.6% mean precision
- Gradual drift: 100% detection in the controlled test battery
- Stuck sensor: 100% detection in the controlled test battery

Important:
These are controlled synthetic validation results, not universal industrial accuracy claims.
Threshold calibration on representative customer data is recommended for domain-specific
or high-stakes deployment.