BULK Exchange Margin: Portfolio Margin Cuts Required Collateral by Up to 70% on Hedged Positions

BULK Exchange uses portfolio margin by default. Instead of evaluating each position’s risk independently, the risk engine evaluates your entire portfolio as a single unit — accounting for correlations between positions. Hedged portfolios use significantly less margin than unhedged ones.

Documented efficiency claim: Up to 70% margin reduction on hedged portfolios.

Portfolio Margin: How It Actually Works

On most exchanges (and on Hyperliquid), margin is calculated per position. Long BTC requires X margin. Short ETH requires Y margin. Total margin = X + Y. Positions don’t interact.

On BULK Exchange, the risk engine computes a correlation-adjusted effective notional for the entire portfolio. Positions that offset each other’s risk are treated as a combined position — which requires less margin than the sum of their parts.

The 9-Regime HMM

The underlying model is a 9-regime Hidden Markov Model:

3 market regimes: Bearish × Neutral × Bullish 3 volatility regimes: Low × Medium × High

The model classifies each market into one of the nine regime combinations. Each combination has a different lambda surface — the three-dimensional function mapping leverage, market impact, and volatility regime to a margin requirement.

No discrete tier jumps. Margin requirements use bilinear interpolation between regime states. There are no cliff edges where crossing a threshold suddenly doubles your margin requirement.

The Correlation Offset

The practical result of portfolio margin: positions that hedge each other require less total margin.

Example — BTC/ETH hedged pair (correlation ~0.85):

The exact BULK requirement depends on current regime classification and the correlation coefficient at the time of the trade. At 70% maximum efficiency, the hedged pair requires ~30% of what per-position margin would demand.

Freed capital = deployable in additional trades, held as buffer against drawdown, or withdrawn.

The Formula: SPAN, Ported Onchain

BULK’s July 16, 2026 “Engine Room” post is the first time BULK has published the actual aggregation math behind portfolio margin — and it confirms the model is explicitly modeled on CME’s SPAN methodology (the standard institutional exchanges like CME, OKX, and Bybit use), not a bespoke onchain invention.

The engine prices margin in two steps:

Step 1 — standalone margin per position. Each position gets its own margin the way a per-position exchange would compute it: Mi = notional × rate. Call them M1, M2, M3.

Step 2 — correlation-weighted aggregation, not addition:

Mp = √( ΣMi² + ΣΣ 2·Mi·Mj·ρij )

ρ (rho) is the pairwise correlation between positions. Two longs in correlated assets reinforce each other and the total stays high. A long and a short in correlated assets partially cancel, and the total drops. Summing margins (M1 + M2 + ...) assumes every position blows up in the same direction at once — the correlation formula asks how likely that actually is.

Worked example — $100K long BTC / $100K short ETH, BTC/ETH correlation 0.81:

The benefit scales with how hedged the book actually is — a single naked long saves almost nothing under this formula, by design.

Regime-break protection: measured correlation isn’t fed into the formula raw. BULK blends the measured ρ with a worst-case ρ, so the offset a hedged book is granted is the one that survives a correlation breakdown in a crisis, not the one from a calm month. This is the same regime-sensitivity the 9-regime HMM classifier below is built to capture — the HMM estimates which correlation regime the market is in; the SPAN-style formula is how that regime-dependent correlation gets turned into a margin number.

Margin rates themselves (the 2% floor in the example, hard caps, leverage limits) are model outputs — a function of volatility regime, the portfolio’s own leverage, and the depth-adjusted cost of force-unwinding the position into the live order book — not a static percentage table.

How this compares to institutional CEX portfolio margin: CME pioneered this exact idea with SPAN in 1988; OKX and Bybit run equivalent worst-case stress grids for their portfolio margin programs today. The difference is access — CEX portfolio margin is gated behind volume thresholds, signed agreements, and VIP tiers. BULK’s version is the only margin engine on the venue, live for every account from the first dollar, and — because BULK is a public state machine — the aggregation formula and the correlation inputs are things you can verify rather than take on faith.

Source: BULK Exchange — “The Engine Room”, July 16, 2026.

Sub-Accounts as Isolation Boundaries

Each sub-account on BULK Exchange runs a completely independent portfolio margin evaluation.

What this means in practice:

• A leveraged long BTC position in sub-account 1 doesn’t affect margin in sub-account 2
• A liquidation event in sub-account 3 doesn’t touch sub-accounts 1 and 2
• You can run completely separate risk profiles across strategies without one strategy’s performance affecting another

This is the institutional use case: separate accounts for different strategies, each with independent risk, all under one master wallet and aggregating volume for fee tier purposes.

Isolated Margin: When to Use It

Isolated margin routes a specific order to a per-instrument isolated account instead of the main portfolio.

How to activate: Add i=true to any order. An isolated account is created automatically on first use. One instrument per isolated account.

Use cases:

• High-conviction directional trade where you want to cap maximum loss
• Experimental positions on volatile assets (memecoins via BIP-1, new listings)
• Positions you don’t want affecting your main portfolio margin
• Teaching yourself BULK’s mechanics without risking your main account

How to size isolated positions: The maximum you can lose equals the capital in the isolated account. Size accordingly.

Portfolio Margin vs. Isolated Margin: The Decision

The Lambda Surface

The lambda surface is the core mathematical object of BULK’s portfolio margin engine. It’s a 3D function:

Inputs:

• Leverage (position size / account equity)
• Market impact (position size relative to ADV)
• Volatility regime (from the HMM classifier)

Output: Margin requirement as a continuous function of these three inputs.

Why continuous matters: most exchanges use tier tables. If your leverage crosses a tier boundary, margin jumps discretely. BULK’s bilinear interpolation means small changes in leverage produce small, proportional changes in margin — no cliff edges.

Cascade Adjustment

The risk engine runs a 2-round iteration to account for liquidation cascade effects.

The problem without cascade adjustment: A position looks well-margined in isolation, but closing it (due to margin breach) could destabilize other correlated positions, requiring further liquidations.

BULK’s solution: Before flagging a liquidation, the engine estimates the price impact of liquidating the position and recalculates whether the portfolio would still be solvent after the cascade. If so, the cascade effect is acceptable. If not, the engine may act earlier to prevent the cascade.

This is the “positions appearing well-margined in isolation but causing systemic risk” problem, directly addressed per the architecture documentation.

Comparing to Hyperliquid’s Margin

Frequently Asked Questions

What is portfolio margin on BULK Exchange? Portfolio margin evaluates your entire account as a single risk unit, using correlation between positions to reduce the total margin requirement for hedged portfolios. BULK claims up to 70% margin efficiency on correlated hedged positions vs. per-position margin.

How does BULK Exchange calculate margin? BULK uses a 9-regime Hidden Markov Model (3 market regimes × 3 volatility regimes) to classify each market. A 3D lambda surface maps leverage, market impact, and volatility regime to a continuous margin requirement with no discrete tier jumps. The regime-dependent correlations that come out of that model feed a published SPAN-style aggregation formula, Mp = √(ΣMi² + ΣΣ2·Mi·Mj·ρij), that combines per-position margins into a single portfolio requirement.

Is BULK’s portfolio margin the same idea as CME SPAN or OKX/Bybit portfolio margin? Yes — BULK describes its engine as SPAN-inspired, the same margin-the-portfolio-not-the-positions idea CME shipped in 1988 and that OKX and Bybit run today. The difference is access: CEX portfolio margin is gated behind volume thresholds and VIP tiers, while BULK’s is the only margin engine on the venue and applies to every account from the first dollar — with the aggregation math itself published rather than a closed backend.

What is isolated margin on BULK Exchange? Adding i=true to any order routes it to a per-instrument isolated account. Isolated positions have a fixed maximum loss (the capital in the isolated account) and don’t affect the main portfolio margin.

How much more capital efficient is BULK’s portfolio margin vs. Hyperliquid? BULK documents up to 70% margin efficiency on hedged portfolios. Hyperliquid uses per-position tiered margin with no correlation adjustment. On a hedged BTC/ETH pair, this difference can free 50–70% of capital that would otherwise be locked as margin on Hyperliquid.

Back to cluster hub: Complete BULK Exchange Trading Guide

Also in this cluster:

• Liquidations: The Optimizer That Preserves Hedges
• Order Types: Market, Limit, ALO, IOC
• Sub-Accounts: 64 Independent Strategies
• Margin Calculator

Related: BULK vs Hyperliquid: Margin Comparison — 70% efficiency vs per-position tiered · Glossary: Portfolio Margin, HMM, Lambda Surface, Isolated Margin

Try portfolio margin live → early.bulk.trade

• What Are Perpetual Futures?
• How Crypto Liquidations Work Last updated: June 3, 2026. Source: BULK Exchange architecture documentation