House margin — where the add-ons live

When a prime broker advertises “risk-based margin financing,” this is the machinery behind it. The base is a house scenario grid — usually wider and more granular than SPAN, sometimes with a historical or Monte Carlo VaR/ES overlay — and the requirement starts as the worst-case loss over that grid.

But the base number is commoditized. The add-ons are where houses differentiate, roughly in order of how often they save you:

• Concentration — position size relative to average daily volume. If liquidation takes eight days instead of one, your effective horizon is eight days, and risk scales roughly with √t. Usually a scaling factor on the base keyed to a days-to-liquidate bucket.
• Liquidity / bid-ask — the cost of crossing the spread on the way out, scaled by size. Distinct from concentration: one is how long, the other is how much worse than mid.
• Gap risk — the grid assumes continuous price paths. Single names gap on earnings; crypto gaps on exchange failures and weekend news. A separate charge for jumps, especially on short options and levered books.
• Wrong-way risk— the client's exposure grows exactly when their creditworthiness falls, or the collateral correlates with the position. A fund posting a token as collateral against a long position in a correlated token is the textbook case. Familiar territory if you've seen CVA and counterparty risk.
• Stress add-on — named historical or hypothetical scenarios (March 2020, LTCM, an exchange collapse, a stablecoin depeg) applied on top, charging only where the stress loss exceeds what the grid already covered.
• Correlation / decorrelation add-on — a charge for portfolios whose margin benefit leans heavily on assumed offsets, because those offsets fail exactly when you need them (Module 5 is entirely about this).