When I was appointed CEO of a new FX broker, I had a plan.
I came from some of the most complex, most regulated corners of the market, I thought to myself. I had also worked with spot FX in hedge funds, interbank feeds, and liquidity providers. Fragmentation didn’t scare me. Oh, no, not me!
What this space needed, I believed, was 360-degree transparency. Clients would be skeptical, at first. Then they would start understanding. Then they would start flying, slowly towards the light. Then in swarms. The industry would recognize the winning pattern and follow, too.
So… let’s revolutionize this market from within!
My incremental growth plan was very solid, I could even call it great. I had laid it out, and the investors had agreed to it excitedly.
But my revolution came to an abrupt end when the investors faced what the plan actually required: a patient build-up and some light investment.
“Our peers went from X to Y in 6 months.” I heard them say at once.
“But... but...”
I would choke on my next words.
I wanted to say “... look at what they have done to so many of their clients...”
“...that is not a long-term business...”
“…you agreed to the plan. I laid it all out.”
I knew there was no point in saying any of it. The revolution was strangled inside me. I had been naive on this occasion. But I WASN’T naive, I had been in the wider industry for too long for that.
When you buy EURUSD with a forex broker, your order doesn’t go to any exchange. There is no single exchange, in fact. You are almost certainly not going to be matched with a bank. (Yes, I can hear some of you pushing back already. We’ll get there.) Who “takes” your order will depend on which of three buckets you fall into. You have very little control, and no visibility into which pile they threw your ticket onto. You may get a helpful line somewhere in the disclosures: “We may be on the other side of your trade.”1 Go do with that line what you want… Most clients don’t know, or don’t care to know, which bucket they’re in. Yet this is one of the most consequential questions in trading retail FX.
Foreign to any Exchange
Before I describe the three buckets, let me emphasize a point. In foreign exchange trading there is no centralized book, no single exchange, no universal “prevailing price” that everybody is entitled to trade at. Exchange-like entities (ECNs) exist, yes — but they are fragments of this market too. No government or organization supervises the entire space. Country by country, the rules are different, and the “regulators” come in all shapes and sizes — different flavors of strictness and legal oversight. Not only that, you aren’t actually buying the underlying currencies in the first place. Contract For Difference means that you and the party on the other side of your trade profit or lose from the difference in price. You click, and sweat through the roller-coaster, but actually hold no Euros.
What you do hold is a risk position, and with that you hand the opposite risk to the broker. When you buy EURUSD, your broker takes on exposure equivalent to a short EURUSD position, before any offset. Why? Well, because if EUR goes up vs USD, they will have to pay you. Let’s understand the options the broker has at their disposal for dealing with this risk, using the three buckets I promised earlier2:
A-book: The I-wash-my-hands bucket
The broker passes your order straight through to a liquidity provider and lets them deal with the risk. They take their cut — a commission, a markup — and sleep tight. Your gain no longer comes at their expense, so your incentives are generally most aligned here. However, there is still a caveat: send too much “toxic flow”3 to the liquidity provider, and they may take action against the broker. The terms worsen, all the way to cutting the service entirely. So even here, it is not quite true that you making money is neutral to the broker. But at least it is the least scary of the three misaligned models.
B-book: “All you can eat”
Your position gets internalized. Which is a fancy way of saying the broker eats that risk because they deem that you and your kind are likely unsophisticated. Which is a fancy way of saying that you are part of the st...d cohort. No offense. And no order, no position goes anywhere.
There may be a hedging mechanism in the background, watching the book’s net exposure against company risk limits and regulatory constraints. Cross a threshold, and some risk gets hedged. But that is a separate mechanism altogether.
From your, the trader’s, perspective, the arrangement starts to look suspiciously like that slot machine in Vegas.
And of course, the incentive structure here is really ugly. If you win, the house doesn’t. But the house also makes money on the machine maintenance, drinks, snacks...
C-book: The undecided.
If the broker doesn’t yet know who you are, or your trading shows characteristics they aren’t comfortable taking entirely onto their own book, they may split the risk. Pass some to a liquidity provider, keep the rest. Which bucket you land in is not random, though. It is the broker’s running verdict on you — informed or not. Start winning consistently and things will start to change for you. First unnoticed, then maybe blatantly. There is a reason an alias for this bucket is the Hybrid. A pinch of A with a spoonful of B.
The onion problem – LPs all the way down
As we said, all the risk your broker doesn’t want goes to liquidity providers. And those liquidity providers may have their own liquidity providers. A game of hot potato, all the way upstream. You get the picture. Behind a layer or two of liquidity providers you may find another dealer, a prime-of-prime giving access, or a prime broker providing the credit. Then there are aggregators combining price feeds, exchange-like venues matching orders, bank feeds, non-bank market makers…4 Some make an impressive list of names in a presentation. But “we connect to major banks” tells you remarkably little about how many mouths your trade has to feed on the way. These layers do provide services — access, credit, technology, liquidity — and they all expect to be paid. And there may not even be a bank at the center of this onion. It might as well be another market maker, doing precisely what your broker decided not to – eat your risk for profits.
Scam?
Does it sound an awful lot like a scam? I hear you, but consider liquidity providers for a second as just market makers. They quote both sides, earn the spread, carry inventory, and try to avoid being picked off by traders who know more than they do. Exactly the mechanism I described in my earlier piece. And how about Payment for Order Flow (PFOF)? In US equities, some market makers pay brokers for the right to execute retail orders. That flow is worth buying because it is generally less likely to pick them off.5 Different machinery from CFD B-booking, but the appetite for uninformed flow is a familiar pattern. I’ll get into PFOF properly in a future piece.
Even the term “toxic flow” has made it into the academic literature as if taken straight out of the mouth of the liquidity providers.6 So similar mechanisms are everywhere. Your broker’s A-book may simply be somebody else’s B-book.
Unfair?
“Yes, ok, but is it fair?” you may ask. Well, in FX, every price depends on who is asking. This goes all the way up. Big institutions using the same wholesale platform can have different prices, different available liquidity, and different counterparties willing to trade with them. Your credit, your relationships, and how much the maker likes your flow all matter.7 “Interbank access” sounds rather more democratic than it actually is.
Then there is last look. The provider shows you a price. You send the order. They get a short window to check it and decide whether to accept or reject. You said yes; they still get to say no.⁹ And if they do, you are back shopping for a price while the market has been moving without you. The tighter spread looked lovely, though, if only it had been real for you.
The stated purpose of last look is, of course, to “check the request’s validity and whether the price is still valid”. So says the FX Global Code. The voluntary (!) FX Global Code.8 A world of stated principles…
An alternate universe
So what world is it then? In my mind it is an alternate universe. In this universe, instead of a fierce competition between trading platforms, you have a quasi-monopoly of a single provider — MT4/MT5 dominate the landscape.¹⁰ Instead of an arms race between professionals, you have an arms race between brokers, and some well-positioned clients: the ones who learned to read a stale quote faster than the broker’s bridge can update it.
Here you have a menu of jurisdictions — pick the one that suits your business model. Instead of a public order book you can check, the broker marks its own prices for you — in a process I playfully call marking to marketing. Instead of exchange fees, a markup at every layer of the chain, all happening behind a comfortingly familiar screen. And instead of a market maker widening the spread to avoid informed flow, the broker decides something far more fundamental: how much of your risk it wants to keep. That is the alphabet. A, B, C.
The result: you are trying to find an edge that survives not only the market, but also the opaque terms you have been asked to sign — and whatever happens upstream. You can spend months studying EURUSD without understanding the particular version of EURUSD you are actually allowed to trade.
And the scoreboard of this universe is published in the brokers’ own risk warnings: most retail accounts lose money. They are forced to publish this.9 I wonder how many people it would stop in Vegas, though, to see the gambling statistics on a big screen. I think we all have an intuition about the answer.
My take on fairness? Tell traders whether they are in A, B or C-book, and when you switch them. Tell them about their latency and slippage settings, and how those differ across client groups. Inform them about the LPs. Show them how well you executed their orders against meaningful cross-industry benchmarks. Tell them about their real prospects. Explain how this setup differs from exchange-traded markets — i.e. a regulated broker does not magically turn a bilateral contract into an exchange order.
“Most of them don’t care,” you say? They do. They are not stupid. Far from it. And they are learning. Give them something more useful than “we may be on the other side of your trade.” Just be transparent.
That was the plan I laid out to my investors. I am holding on to the hope that somebody in the industry takes it up some day.
Here I mean retail OTC currency CFDs and similar non-deliverable rolling FX products. Deliverable currencies and exchange-traded currency futures are different arrangements. In the common principal model, your contract remains with the broker even when they pass the risk to an LP. See ESMA's CFD Q&A, Section 2.
A/B/C are industry labels, not a standardized regulatory alphabet. Here, A-book means immediate, full pass-through; B-book means internally retained client risk; C-book means a deliberate split. Aggregate risk hedges can still sit over B-book positions. ESMA’s broader “hybrid” category includes threshold-based hedging and hedging selected clients. See ESMA’s CFD Q&A, Section 2.
LPs assess flow using markout: how the market price moves after a fill, measured against the execution price. Flow that consistently moves against them shortly after execution is costly to service. Your eventual profit and their short-term markout are different measurements.
These firms perform different jobs. Prime brokers supply credit and access; aggregators combine quotes; venues match orders; dealers take positions. They aren’t necessarily successive counterparties to the same trade. The onion includes infrastructure as well as risk-taking. See BIS, FX trade execution: complex and highly fragmented.
The SEC’s staff describes retail equity flow as attractive because it is generally less informed about short-term price movements. Equity market makers also internalize orders; PFOF is the payment for routing that flow, rather than a CFD book classification. See SEC staff’s market-structure memo, Section III.
See Easley, López de Prado and O’Hara’s Flow Toxicity and Liquidity in a High-frequency World (2012). For an FX-specific example, Cartea, Duran-Martin and Sánchez-Betancourt’s Detecting Toxic Flow models a broker’s decision to internalize or externalize client trades.
For wholesale pricing mechanics, see Graham Capital’s Execution in Spot FX. Retail brokers subject to MiFID face a different standard: ESMA rejects different prices, spreads or unnecessary delays for similar orders based on client profitability or the broker’s hedging arrangements. See Section 9, paragraph 37. Enforcement rests with national authorities. A rule on paper and a trader’s ability to get it enforced are, of course, two different things.
Principle 17 of the FX Global Code describes last look as a validity and/or price check, and calls for disclosure of its use. The Code concerns the wholesale FX market and is voluntary: it creates no legal or regulatory obligations itself. Local laws and contractual obligations still apply.
For example, UK rules require provider-specific loss percentages. It counts accounts, not individual trades, and may cover several speculative products rather than FX alone.