Every trade your clients place lands on your book. How brokerages measure exposure, decide what to hedge, and build a dealing desk that survives the news spikes that break weak risk systems.
BridgingFX editorial · 2026-09-25
Ask a new brokerage founder about their plans and you'll hear about platforms, liquidity, and marketing. Ask what happens when a hundred clients go long gold into a Fed announcement and you'll often get silence. Risk management is the least glamorous part of running a brokerage and the one that decides whether the business survives its first volatility shock. This is the foundational tour: how exposure works, what hedging actually means in practice, and how a dealing desk is supposed to function.
Every client trade creates a position on the broker's book. If clients are collectively long 500 lots of EUR/USD and the broker hasn't offset that exposure, the broker is short 500 lots — it profits when clients lose and loses when clients win. This net exposure, aggregated across every symbol and client, is the single number the dealing desk watches. Everything in broker risk management is about measuring that number accurately and deciding what to do about it.
The key subtlety: gross exposure and net exposure tell different stories. A book with 1,000 lots long and 950 lots short has 1,950 lots of gross flow but only 50 lots of net exposure. Client flows naturally offset each other — this 'internalisation' is free risk reduction, and it's why brokers with large, diverse client bases can run leaner hedging operations than small ones. Scale doesn't just cut unit costs; it literally reduces risk.
The industry shorthand: A-book means passing client flow to liquidity providers (the broker earns spread/commission, takes no market risk); B-book means internalising it (the broker takes the other side). The caricature — A-book good, B-book evil — misses how the business actually works. Almost every brokerage runs a hybrid: toxic or large flow gets hedged (A-booked), while small, diversified retail flow is internalised, because hedging every micro-lot would cost more in spread than the risk is worth.
The real risk-management question isn't which book you're on — it's whether your hedging decisions are systematic or hopeful. A disciplined hybrid books flow by rule: client profitability profiles, position sizes, concentration limits, and market conditions decide what gets hedged, automatically. An undisciplined one B-books everything and prays. The first is a business; the second is a casino that doesn't know the odds.
Hedging means taking an offsetting position with a liquidity provider — if clients are net long gold, the broker buys gold from its LP so a price spike doesn't create a loss. But hedging isn't free: every hedge crosses the LP's spread, and hedging too eagerly bleeds the dealing profit that internalisation earned. The desk's job is optimisation under uncertainty: hedge enough to survive the tail events, internalise enough to stay profitable, and never let a single symbol's exposure exceed the limits the risk policy sets.
News events are the stress test. Exposure that looks manageable in calm markets can become lethal in seconds when price gaps — hedges placed after the move are hedges placed too late. Serious desks run pre-news procedures: reduced leverage, widened internal alerts, and exposure caps tightened ahead of known events. The brokers that blow up on news day are almost never surprised by the news; they're surprised by their own book.
A dealing desk isn't necessarily people shouting at screens — for most modern brokerages it's a function: monitoring systems, limit frameworks, and escalation procedures, with humans overseeing the exceptions. The components that matter: real-time exposure aggregation across all platforms and symbols, automated alerts at exposure thresholds, STP/hedging execution connectivity that works when markets are moving fastest, and a written risk policy that says who can override what.
Technology choices compound here. A bridge that adds 300 milliseconds to hedge execution is a minor annoyance in calm markets and a serious liability during a spike. Reporting that reconciles hourly instead of in real time means the desk is always managing the last hour's risk, not this minute's. When evaluating bridge and risk-system vendors, the question isn't the feature list — it's behaviour under the load profile of your worst day.
The post-mortems rhyme. A broker grows fast on marketing, internalises everything because hedging 'costs margin', concentrates exposure in a trending symbol, and then a gap move — a central bank surprise, a weekend event — turns the book into a loss the capital base can't absorb. Client profits are the broker's losses, and they're owed in full. Undercapitalised B-books don't survive their first real shock; that's the entire history of retail broker failures in one paragraph.
The founders who survive build risk management before they need it: capital buffers sized for gap scenarios, systematic hedging rules, exposure limits enforced by systems rather than intentions, and a risk function with genuine authority to say no to the sales team. None of this appears in a launch checklist's marketing section. All of it determines whether there's a second year.
Broker risk management is the discipline of measuring net exposure and hedging it systematically — not the choice between A-book and B-book, but the rules that govern the hybrid every real brokerage runs. Internalisation profits from diversified flow; hedging protects against concentration and gaps. The brokers that fail treat risk as overhead. The ones that last treat it as the business itself.
Let's scope it — honestly, and in writing.
