The Proof.
We ran SIGMA Engine retroactively on 8 major sovereign crises using historical parameters. Below are the scores the model would have produced before each event. No hindsight. No adjustments. Same engine. Same math.
Bloomberg Terminal does not publish its historical accuracy. We do.
Every bubble = one monitored entity. Entities above the diagonal line = SIGMA sees more risk than markets price. Current biggest divergence: 🇫🇷 France — 31 points unpriced.
Historical scores use research-backed parameter approximations for each crisis period. Same engine, same math as live scores.
Market-implied risk = approximate proxy derived from published yield spreads and CDS levels. Not investment advice.
Divergence = Noosphere Score minus market-implied risk. Positive = model sees more danger than markets price.
Every live prediction. Timestamped. Brier scored.
Bloomberg doesn't publish its track record. This is ours.
View Full Ledger →Get Alerts