The 25 controls of the risk engine

Every trade on Hypertrade passes a deterministic contract of 25 controls. Safety controls can veto or resize; evidence controls rank opportunities without pretending to be safety rules.

Updated 2026-07-21

The risk engine is deterministic code that sits between a proposal and the exchange. Each rule ID has exactly one meaning and one owner. Safety checks may approve / resize / veto / delay; allocator checks are explicitly advisory and can never silently increase leverage.

RuleWhat it protects
R-01 · Protective stopEvery entry requires a valid protective-side stop at a sane distance. Enforcement: hard veto.
R-02 · Universe allowlistOnly a market permitted by the desk mandate may be traded. Enforcement: hard veto.
R-03 · Leverage ceilingLeverage is capped by the strictest mandate, platform, and venue-asset limit. Enforcement: approve or resize.
R-04 · Position directionAn order may not accidentally flip, double, or conflict with existing exposure. Enforcement: fresh execution preflight.
R-05 · Risk per tradeProjected loss to the protective stop is resized to the desk's per-trade budget. Enforcement: approve or resize.
R-06 · Venue collateralThe destination perp DEX must have readable, usable collateral before an entry can route. Enforcement: hard veto.
R-07 · Daily loss limitNew risk stops after the desk reaches its daily realized-loss boundary. Enforcement: hard veto.
R-08 · Data freshnessA decision cannot execute from stale or incomplete decision-grade market data. Enforcement: hard veto.
R-09 · Drawdown auto-pauseThe desk pauses when equity drawdown reaches the configured boundary. Enforcement: hard veto.
R-10 · Strategy cost admissionA strategy applies its frozen cost model before proposal; no generic fixed-spread profile gate may replace it. Enforcement: fresh execution preflight.
R-11 · Size-aware liquidityVisible depth must support the final order size without unacceptable market impact. Enforcement: fresh execution preflight.
R-12 · Venue order boundsFinal notional and size must satisfy venue minimums, precision, and platform ceilings. Enforcement: hard veto.
R-13 · Initial-margin feasibilityThe order is resized to initial margin that the destination account can actually open. Enforcement: approve or resize.
R-14 · Liquidation and maintenance bufferIsolated positions keep the stop before liquidation; cross positions must retain stressed maintenance headroom. Enforcement: approve or resize.
R-15 · One net position per symbolManaged entries cannot stack. Repeated same-direction manual tickets add to the wallet's one net position and one canonical trade record; opposite orders are netting, not separate hedges. Enforcement: hard veto.
R-16 · Re-entry cooldownA recently exited market waits for a new independent strategy bar before it may re-enter. Enforcement: temporary delay.
R-17 · Factor exposureNet and gross factor exposure are resized inside whole-wallet stress ceilings. Enforcement: approve or resize.
R-18 · Account healthNo new risk is allowed when maintenance margin is already too close to account equity. Enforcement: hard veto.
R-19 · Alpha evidenceStrategy strength is an allocator input, not an uncalibrated probability or independent safety veto. Enforcement: allocator score only.
R-20 · Funding and carryExpected funding and carry are charged to net expectancy instead of applying a fixed directional veto. Enforcement: allocator score only.
R-21 · Plan capacityPer-position notional is resized to the user's subscription-plan allowance. Enforcement: approve or resize.
R-22 · Projected portfolio stressPost-trade mark-to-stop risk, gap loss, direction, and factor concentration must fit local and wallet-wide budgets. Enforcement: approve or resize.
R-23 · Macro-event delayFresh entries wait through configured tier-one macro release windows. Enforcement: temporary delay.
R-24 · Initial-margin headroomAdditional initial margin must fit candidate-DEX account value minus margin already in use and the strategy/profile reserve. Transfer-withdrawable collateral is tracked separately. Enforcement: approve or resize.
R-25 · Capital efficiencyCapital efficiency ranks feasible opportunities but never raises leverage or vetoes an otherwise valid trade. Enforcement: allocator score only.

Properties worth knowing

  • Deterministic: same inputs, same verdict, every time. No AI variance in the safety layer.
  • Independent: the AI cannot override a hard safety check.
  • Transparent: every resize, delay, and veto is shown with its canonical rule ID.
  • Fail-closed at live boundaries: unreadable venue, account, protection, or execution data cannot open new risk.
  • No false precision: alpha strength, funding, and capital efficiency rank trades through net expectancy; they are not advertised as calibrated win probabilities.
This inventory is generated from the same canonical RuleSpec registry consumed by the product and semantic contract tests.