Aggregated risk intelligence from institutions across the network, privacy-preserving, so you can see when your book is above or below the network — without exposing anyone's positions.
Subscribe to the oracle →When markets move, the most important question is often: "Is my book an outlier, or is everyone feeling this?" The Risk Oracle answers that question without requiring anyone to share their book. Each participant's risk signals are aggregated with differential privacy noise — a mathematical technique that provides a provable privacy guarantee. What you see is the network view: regime classification (calm, stress, crisis), aggregate flow patterns, and concentration trends. No individual book is ever visible.
Participants contribute risk signals
Each institution's system publishes anonymized risk signals — aggregate DV01 by bucket, concentration indices, flow direction — with differential privacy noise applied locally before transmission.
Signals are aggregated with DP noise
The oracle combines signals using an ε-differential privacy mechanism. The noise budget ensures that no individual participant's contribution can be inferred from the output, even by an adversary with full knowledge of the aggregate.
You receive the network view
The oracle classifies the current market regime, shows where aggregate flows are concentrated, and flags when your book's sensitivity profile is above or below the network median. k≥3 anonymity is enforced on every data cell.
Market regime
Calm / stress / crisis classification derived from aggregate volatility, flow, and credit signals across all participants.
Aggregate DV01 flow
Where is the network adding or reducing rate sensitivity? By asset class, by tenor bucket.
HHI concentration
System-wide Herfindahl-Hirschman Index — is market concentration increasing or dispersing?
Proof-of-inferiority benchmark
Is your hedge ratio above or below the network median for your segment? Used in the benchmark scorecard.
ε-Differential Privacy
Statistically provable privacy guarantee. Calibrated noise is added to every contribution before aggregation.
k≥3 Anonymity
No data cell is published with fewer than 3 contributors. Small cohorts are suppressed.
Advisory only
Oracle output is informational. It does not constitute investment advice or a recommendation to trade.
Opt-in contribution
You control which signals your institution contributes and which oracle feeds you subscribe to.
Can anyone identify my book from the oracle output?
No. Differential privacy (ε-DP) adds calibrated noise to each contributor's signal before aggregation. Even with full knowledge of the aggregate output, it is mathematically infeasible to determine any individual participant's contribution. The k≥3 anonymity rule additionally ensures no data cell is ever built from fewer than three contributors.
Is this investment advice?
No. The oracle is advisory intelligence — it shows you where the network stands relative to your book. It does not recommend specific trades, does not constitute investment advice, and does not replace your own risk management judgment.
What data feeds can I subscribe to?
Available feeds include: regime classification (calm/stress/crisis), aggregate DV01 flow by asset class, HHI concentration trends, counterparty credit signals, and liquidity depth proxies. You choose which feeds your agents can read and how frequently they refresh.