Institutional portfolio risk & margin calculation
A real-time, collateral-aware risk and margin engine for digital-asset trading infrastructure — built for exchanges, brokers, prime brokers and clearing venues.
BTC
ETH · majors
Modern digital-asset portfolios outrun static risk controls
Customer books now blend spot, perpetuals, futures, options and volatile collateral — while margin logic stays product-based, static, or too slow for live trading.
Mixed, nonlinear books
Spot, perps, futures and options no longer behave like isolated line items.
Collateral that moves
BTC, ETH and stablecoins carry their own risk — yet are often netted as static value.
Linear approximations
Notional percentages and Greeks-only proxies miss offsets, concentration and stress.
Too slow for live flow
Pre-trade checks and liquidation can't wait on a stale or partial recalculation cycle.
What is the customer's required margin right now?
How much available margin remains after positions, collateral and market risk?
How much more can they trade before breaching the portfolio constraint?
How does that change under current vs. stress views?
How are option exposures valued if the account becomes distressed?
One portfolio-aware layer, not a stack of silos
CloudRisk Diversity evaluates the whole book as a single integrated risk object — returning margin, capacity and stress analytics within a defined low-latency budget.
Close-out & maintenance margin
Liquidation and margin-call levels from one model.
Initial margin
Upfront and ongoing trading buffer above the call level.
Available margin
Resources net of applicable margin, at account level.
Single-order limits
Tradeable long / short capacity by instrument & direction.
VaR & expected shortfall
Portfolio diagnostics where enabled.
Current & stress PnL vectors
Blended, stress-aware margin measures.
Full revaluation, expressed as PnL vectors
Every instrument is revalued under shocked scenarios and represented as a PnL vector. Portfolio risk is the aggregation of those vectors — capturing offsets, concentration and nonlinearity that static percentages cannot.
- —Fixed percentages by product
- —Blind to portfolio composition & offsets
- —Ignores collateral volatility
- —Greeks-only / linear proxies for options
- —Over-restricts trading or understates risk
- →Risk computed from the actual portfolio
- →Captures offsets, concentration & netting
- →Collateral flows through the same framework
- →Options fully revalued — no linear shortcut
- →Current & stress views blended via alpha
From portfolio scenarios to margin, capacity and headroom
Collateral is risk — not a static deduction
Collateral balances are represented by their economic risk profile, so changes in collateral value flow through the same PnL-vector framework as trading positions.
Volatile by nature
In digital-asset markets, collateral can move as much as the positions it backs.
The static-netting trap
Engines that ignore collateral risk overstate resources exactly when stress hits.
Benefit and risk, together
Outputs reflect both the value of collateral and the risk of holding it.
Turning portfolio risk into a live trading control
For each enabled instrument, the engine solves how much additional long or short quantity the customer can add before reaching the portfolio constraint — portfolio-specific, not a static order-size cap.
Options follow the same principle — unit option PnL vectors scale linearly in contract quantity, so long / short option capacity is solved in the same framework.
- ›Pre-trade risk checks
- ›Available-to-trade displays
- ›Instrument-level capacity controls
- ›Portfolio-aware order validation
- ›Faster flow, less manual intervention
- ›Safer expansion of leveraged products
A multi-exchange real-time feeder, built into the risk core
The feeder connects to exchange websocket streams, normalizes venue-specific messages, enriches them with common metadata, and routes data into pricing, calibration, market-state construction and option-surface updates.
Live websocket ingestion
Direct connections across major crypto venues.
Normalization & enrichment
Venue messages mapped to one instrument vocabulary.
Routed into the risk core
Feeds calibration, pricing and finalized market states.
BTCETH · other major digital assetsOptions fully revalued — a tool for distressed books
Each option contributes a shocked revaluation vector reflecting spot, dividends, curves and Heston parameters, aggregated into the same portfolio outputs. No Greeks-only shortcut.
Heston pricing
Backward-flat parameter selection for arbitrary maturities.
Single & batch portfolio pricing
Values a set of positions from one consistent market state.
Auditable native Greeks
Aggregated by factor — no misleading scalars across currencies.
Risk and liquidation teams can compare immediate liquidation against hedged run-off or hold-to-maturity — instead of treating a forced sale as the only path.
Analytics support default-management decisions — they do not replace the customer's liquidation discretion.
Heavy preparation upstream — deterministic answers on the live path
Ingest · normalize · calibrate · build option surfaces · run historical simulation
Reads only finalized market states · portfolio aggregation · margin & single-order-limit solving
Target calibration cadence ~60s · the live path never consumes a failed or partial calibration — it holds the last finalized state.
Determinism
Every live request is calculated against one defined risk state.
Latency control
Heavy market preparation stays off the live request path.
Operational clarity
Degraded-mode behaviour is easy to monitor, explain and govern.
Built for live trading speed — and horizontal scale
Calibration and simulation are done upstream; each live request simply reads a finalized state and aggregates. Throughput grows by scaling out the live service — without changing methodology or relaxing the latency target. Feeders, calibration, market-state construction and risk calculation all scale independently.
External performance claims to be supported by benchmark methodology, concurrency levels, percentile latency and deployment topology.
Designed for resilient production operation
Core data, messaging, orchestration and secrets-management run in clustered configurations across two regions with multiple data centres each — no critical work pinned to a single process or machine.
Multi-region, multi-DC
Two regions, multiple data centres — geographic and infrastructure redundancy.
Separated services
Calibration and calculation are decoupled; worker-based processing where it fits.
Clustered state
Operational state held in clustered infrastructure; node failures absorbed.
A clear source of truth
If no new state is available, the service holds the last finalized one — never partial.
For institutional deployment, supported by failover behaviour, recovery objectives, monitoring model, alerting procedures and SLA assumptions.
A clean line: CloudRisk calculates, the customer governs
CloudRisk Diversity provides quantitative outputs — margin, available margin, collateral-aware risk, stress analytics and single-order limits. Prudential, legal and default decisions remain with the customer.
- +Portfolio margin & available margin
- +Collateral-aware risk
- +Stress-sensitive analytics
- +Single-order limits
- +Fast, consistent, explainable outputs
- ›Margin policy & minimum floors
- ›Concentration & liquidity overlays
- ›Collateral eligibility & haircut policy
- ›Custody, reconciliation & enforceability
- ›Liquidation & default management
- ›Approval of methodology changes
Evaluate with confidence
The full institutional proposition is supported by an evidence pack — methodology transparency, performance evidence, operating resilience and a clear integration path.
Request a briefing
Tell us about your venue and we'll arrange a technical walkthrough.