Unified portfolio risk graph
Aegis is the risk engine beneath every account on ORBITRA ONE™. It models collateral, volatility, liquidity, concentration, correlation, counterparty exposure, Greeks, liquidation paths and protection policies as one connected graph, and evaluates that graph inside ApexMatch before any order reaches the book. For margin modes as a trader uses them, see risk; this page describes the engine.
How it moves
AEGIS
Unified risk graph
Select a risk dimension
Collateral · Volatility · Liquidity · Concentration · Correlation · Counterparty
A radial exposure graph with the portfolio at its center and six risk dimensions around it: collateral, volatility, liquidity, concentration, correlation and counterparty. Hovering over or focusing a dimension shows how a change in it, such as a deeper collateral haircut or a volatility spike, flows through to margin used and liquidation distance at the center.
The problem
Most trading stacks measure risk in silos. The perpetuals venue knows its margin, the options desk knows its Greeks, the custodian knows the collateral and the bot platform knows nothing at all. Each silo can look healthy while the combined portfolio sits one move away from forced liquidation.
Silos also fail in time. Risk is often checked after the fill, recalculated on a timer and enforced by liquidation engines that act on a single price source. The first moment a trader sees the true exposure is too often the moment it is being closed.
Aegis treats risk as a property of the whole portfolio. It is computed before commitment, recomputed as markets move and unwound by rules published in advance. Offsets are recognized where the margin mode allows, and hidden concentrations are surfaced.
Operating sequence
Every operation reads the same graph, so the figure a trader sees before confirming comes from the rules that later protect the account.
Positions, collateral balances, open orders, agent permissions and protection policies become nodes in the account’s graph. Edges carry correlation, collateral haircuts and counterparty links.
Before confirmation, the proposed order is applied to a copy of the graph. Margin used, liquidation distance, concentration and any policy the order would touch return to the ticket.
When the sequenced order reaches ApexMatch, Aegis evaluates it again, synchronously, against current state. The preview is advisory; the gate is authoritative. A breach is rejected with its reason.
Each Prism update re-prices only the nodes it affects. Exposure, stress results and collateral quality stay current without recomputing the whole portfolio.
Stops, hedges, circuit breakers and agent kill switches fire on the conditions their policies define — before maintenance margin is exhausted, not after.
If maintenance margin is breached, partial liquidation reduces exposure along the account’s precomputed path until margin is restored. Losses beyond the account’s collateral follow the insurance waterfall; auto-deleveraging applies only under published conditions.
Architecture
Aegis is a native market-services module of Orbitra L1. Its components are deterministic: given the same state, parameters and market inputs, every validator reaches the same verdict.
The account-level data model. Nodes are positions, collateral, open orders, agents and policies; edges carry what makes risk non-additive — correlation, haircuts, shared counterparties and hedges.
Signed mark prices and confidence scores from Prism, order-book depth from ApexMatch and implied-volatility surfaces for options. Confidence is an input to the model, not a footnote.
Isolated, cross and portfolio margin run on one graph. Portfolio margin evaluates the account across price and volatility scenarios, recognizing offsets that per-position margin ignores. Options add delta, gamma, vega and theta.
Applies a proposed order to a copy of the graph and returns the post-trade state. The same simulator serves the ticket preview, Cortex agent simulations and the synchronous gate.
Replays the portfolio through regime shocks, correlation breaks and liquidity withdrawal. Results stream to the account and are available through stress and scenario APIs.
Holds user stops, institutional limits and agent policies as versioned state. Every verdict records the policy and parameter versions it was computed under.
Maintains each account’s liquidation path, executes partial liquidation through reduce-only orders and applies the insurance waterfall and auto-deleveraging ranking in a fixed order.
Security and failure control
Aegis controls how risk is measured, limited and unwound; it does not remove market risk. Leverage can magnify losses as well as gains — see trading and leverage risk. Its own failures are designed to err toward protection.
Across the three systems
Orbitra Prime
Trading intelligence
Every order ticket in Orbitra Prime shows the Aegis preview — margin used, liquidation distance and policy impact — before confirmation. The portfolio risk panel is a direct view of the account’s graph, and risk explains the margin modes it supports.
Orbitra L1
Settlement and compute
On Orbitra L1, Aegis sits in the market-services layer beside ApexMatch. It evaluates orders inside the same state transition as ApexMatch, reads Prism market state, and its verdicts are replayed by validators before QSE finalizes them.
Orbitra Realm
Applications and commerce
Realm applications that hold positions or collateral for a user register them in the same graph, so a treasury tool or tokenized-asset vault sits inside the owner’s risk view, not beside it. Agents running in Realm answer to the same kill controls.
Value
Specifications
Terminology