Proof, not promises.

Misprice runs a public proof phase. Here's what it proves — and what it doesn't.

System Status

LIVE

Proof PhaseDay 14 / 28

Uptime

99.7%

Cycles—

Settlements—

Evidence Hashes—

Last Health CheckConnecting...

Live metrics will update automatically during proof phase.

What This Proves

✓The system runs continuously without crashes
✓Data pipelines pull and normalize successfully
✓Evidence chains are unbroken and hash-verifiable
✓Settlement logic waits for gamma (event end)
✓Kill switches block signals when gates fail
✓Daily and weekly reports generate automatically

What This Does NOT Prove

✗That mispricing signals are profitable in live markets
✗That the truth model is calibrated
✗That historical patterns will persist
✗That paper results match live execution

We are proving the system works. We are not claiming it makes money.

Evidence Chain

Every action in Misprice produces a hash. These hashes chain together — if any input changes, the hash changes. You can verify any signal's provenance back to raw market data.

Snapshot → SHA-256(market_state + timestamp)

Signal → SHA-256(inputs + decision + rationale)

Settlement → SHA-256(signal_hash + outcome + gamma_flag)

Evidence Chain

Awaiting live connection

Evidence entries will appear here during proof phase.

Weekly Proof Reports

Every week, we publish:

Settled event count
Kill switch activation rate
Edge quality distribution
Model calibration metrics (predicted vs actual)
System health summary

Reports are timestamped and hash-signed.

Digital Twin Validation

PASS

The Anti-Hallucination Test: We run Misprice against synthetic data with no alpha (Null World). If the system finds edges in random data, it's hallucinating.

Expected Edge Rate

~0%

Observed Edge Rate

< 0.5%

Last run: Pending live data

All metrics are from paper mode. No real capital is at risk. Past paper results are not predictive of future outcomes.

See Architecture How It Works