RiskState
RiskState turns live crypto market data into deterministic exposure limits and leverage caps for traders, allocators, an
The Problem
Crypto traders, funds, and autonomous trading agents deploy capital without a consistent, auditable way to size positions or cap leverage against changing market conditions. Discretionary traders rely on emotional sizing, while agents and bots typically act on signals without a binding pre-trade check on exposure. This leaves portfolios exposed to deep drawdowns during regime shifts like panics, euphoria spikes, or DeFi liquidation events, since there is no shared, deterministic layer enforcing limits before capital is committed. The page frames this as a lack of a 'risk layer' sitting between market state and execution, across spot, perpetuals, and DeFi borrowing.
The Solution
RiskState is a deterministic engine that ingests 9+ live market feeds (price, derivatives, on-chain, macro, DeFi, sentiment) and converts them into a composite score, market regime (e.g. PANIC, EUPHORIA, TREND), and cycle phase. This feeds a policy engine that outputs risk permissions: a 5-level policy (L1 Survival to L5 Expansion), max exposure fraction, leverage caps, and explicit allowed/blocked actions (e.g. DCA, LIMIT_LONG_CONFIRMED vs ALL_IN, LEVERAGE_GT_2X). Outputs are delivered via a REST API or dashboard, with per-API-key mandate enforcement (max gross, leverage cap, blocked protocols, per-asset caps) and a full audit trail including policy_hash, scoring_version, and reason_codes. The engine is described as deterministic and externally mathematically reviewed, using published methods like sigmoids, logistic policy caps, and Fisher z-transform confidence intervals, with findings forward-validated against 600+ live snapshots using permutation tests and bootstrap methods. It explicitly does not generate trading signals, execute trades, or custody assets, positioning itself as an independent risk layer for allocators, AI agents, and traders.
