How institutional credit facilities work in crypto markets, and why undercollateralised lending requires a fundamentally different risk framework than traditional finance.
The capital efficiency problem
For most of crypto's history, lending has meant one thing: post $150 of collateral, borrow $100 of stablecoins, get liquidated the moment the market sneezes. That model is fine for retail DeFi. It's useless for a trading desk.
Professional traders live and die by capital efficiency. A market maker quoting fifteen venues, an arb desk working the basis between a CEX and a DEX, a quant fund running delta-neutral — none of them can afford to park multiples of their strategy size as collateral just to get into the trade. What they need is undercollateralised lending: credit extended on the basis of who they are, how they trade, and how their risk is monitored in real time — not just how much they've pledged.
That's what institutional prime services in digital assets really sell. Everything else — execution, reporting, capital introduction — is built around making that lending responsible.
What "undercollateralised" actually means here
A professional facility looks roughly like this: you post $1M in qualifying collateral and you get buying power of several million to deploy across connected venues. The structural features worth knowing:
- Cross-margining across venues. Your long on Binance and your short on Deribit net into one portfolio view. Offsetting exposures reduce your margin rather than double-charging you.
- Continuous mark-to-market. Collateral and positions get revalued against live order books, not yesterday's close.
- Dynamic limits. Leverage flexes with volatility, concentration, and liquidity. It isn't a sticker on the door.
- Credit-based extension. Beyond collateral, the lender is underwriting you — your strategy, your track record, your drawdown history, your ops.
Same structural logic as traditional prime brokerage. What's different is the market: 24/7, fragmented liquidity, wildly varying settlement finality between a CEX trade and an on-chain swap.
Why legacy risk frameworks don't port over
A traditional prime broker running an equity long-short book has centralised clearing, T+1 settlement, and a regulated sec lending market to lean on. Crypto has none of that.
Markets run around the clock. Liquidity can disappear in minutes when macro turns. Perpetual funding rates flip violently. On-chain liquidations cascade through AMMs in ways that don't exist in TradFi. A trader running a simple basis trade between a CEX perp and a DEX spot is simultaneously exposed to exchange risk, oracle risk, bridge risk, and smart contract risk.
Any serious lending programme in this space has to deal with four things that legacy frameworks just don't touch:
- Venue heterogeneity. Every exchange has its own margin rules, liquidation engine, API quirks, insurance fund. You have to normalise across all of them or you're flying blind.
- Speed. When BTC drops 5% in ninety seconds, your risk monitoring and hedge execution have to happen inside that same window.
- Collateral mobility. Collateral at venue A doesn't instantly cover exposure at venue B. Treasury and settlement logistics are part of the risk model, not an afterthought.
- Protocol risk on DEXs. On a CEX, your counterparty is an institution. On a DEX, it's code. Both need pricing.
Ignore any of these and you end up either over-lending into tail risk you can't see, or under-lending to the point where nobody actually benefits.
The liquidation problem — and why we cap native leverage at 3x
Here's the thing people tend to gloss over: even if your prime broker is running sophisticated portfolio-level risk, the exchanges where your positions actually sit are running their own liquidation engines. And those engines don't care about your portfolio view.
If you're running 10x, 20x, 50x leverage natively on a single exchange — which plenty of venues will happily let you do — you're one wick away from being auto-liquidated by that venue's risk engine, regardless of how hedged your overall book is. The exchange sees your position in isolation. A brief spike, a thin order book moment, a funding rate reset at the wrong second, and you're out. No recovery, no appeal, no explanation — just a liquidation notice and a hole in your P&L.
This is why Bequant caps native exchange-level leverage at 3x, even when the overall portfolio facility allows more. You get the capital efficiency of an undercollateralised facility at the portfolio level, but we don't let you put yourself in range of an exchange's liquidation engine on any single venue. The cross-margin benefit sits at the prime broker level, where it's actually managed — not at the venue level, where one bad print can wipe out an otherwise well-hedged position.
It's a boring-sounding rule that saves traders from a very specific, very common disaster.
Where DMA comes in
You can't responsibly extend leverage to someone if you can't see their positions in real time, route their orders with minimal latency, or hedge when limits get approached. That's why execution infrastructure is the actual foundation of the lending product — not an add-on.
Bequant's DMA infrastructure plugs a single account into the major CEXs and, increasingly, into DEX venues through institutional connectors. Orders route directly, without middle layers adding latency or hiding fills. Positions, balances, and open orders stream back into one unified risk ledger.
For lending, that translates into three things:
One view of the book. Every trade on every venue hits the same internal risk ledger the moment it fills. No reconciliation lag between the OKX perp short and the Uniswap spot long.
Consistent collateral accounting. Collateral across exchanges and custody gets marked to the same reference prices, with venue-specific haircuts baked in based on withdrawal liquidity and operational risk.
Hedges execute on the same rails as client orders. When the risk engine decides something needs to happen, it uses the same connectivity, the same latency profile, the same venue relationships.
Without that foundation, leverage is guesswork. With it, it's a product.
RiskQuant: the engine behind the facility
RiskQuant is Bequant's real-time risk engine. It's what makes the lending product defensible. Three things it does continuously:
Portfolio-level margin
Static margin percentages per position are a blunt instrument. RiskQuant calculates required margin against the portfolio's actual risk. A BTC long on Binance hedged by a BTC-PERP short on Deribit gets recognised as a basis trade — not two independent directional bets. That's what makes the leverage economically meaningful. It applies to net risk, not gross notional.
Stress testing
Every portfolio gets revalued continuously against a library of stress scenarios — flash crashes, exchange outages, stablecoin depegs, funding spikes, oracle manipulation, correlated deleveraging events. Margin requirements tighten automatically when the book becomes more sensitive to these, even if the P&L looks fine right now.
Concentration and liquidity haircuts
RiskQuant penalises concentration in illiquid assets, positions that are large relative to venue order book depth, and collateral held in single-venue or single-chain form. $5M of a mid-cap token as collateral gets valued very differently from $5M of BTC split across multiple custodians. That's as it should be.
What you can actually do with it
With DMA providing the rails, RiskQuant running the risk, and native leverage capped at 3x for safety, the strategies that open up are the ones institutional traders actually want to run:
Cross-exchange arbitrage. Price dislocations between CEXs, or between CEX spot and DEX AMM pools, captured with leveraged capital from a single margin account.
Basis and funding trades. Cash-and-carry between spot and perps, or between perp venues, with margin that recognises the hedge instead of punishing it.
Market making at scale. Quote size across multiple venues without posting full collateral at each. Capital stays free for inventory management and adverse selection buffers.
DeFi yield with CEX hedging. Capital into DEX pools or lending protocols, directional exposure hedged on centralised venues, cross-venue margining making the combined trade capital-efficient.
Relative value. Term structure, implied-versus-realised vol, cross-asset spreads — all expressed with leverage calibrated to the actual risk of the position.
In each case, what makes the trade bankable isn't just the leverage. It's the combination: capital efficiency, accurate cross-venue risk measurement, and execution you can trust to enforce the limits.
Prime services: the wider picture
Lending is the headline. The rest of the prime services stack matters more than people give it credit for: allocation of capital across internal strategies and sub-accounts, capital introduction connecting trading talent with allocators, consolidated reporting, treasury across fiat and digital asset rails.
Standalone lending is a facility. Lending inside a prime services offering is infrastructure — the kind of thing you can build a trading business on. The same DMA connectivity that supports leveraged execution also handles sub-account allocation logic and investor-grade reporting. The same risk engine that governs credit limits also powers the reports allocators need to size you.
For traders building something, and for allocators evaluating them, that integration is the whole point. Capital is more productive when execution, risk, financing, and reporting all sit on the same stack.
Bequant provides institutional prime services for digital assets: undercollateralised lending, DMA execution, cross-venue risk management through RiskQuant, allocation of capital, and capital introduction. To talk about a facility, get in touch with the Bequant institutional team.