Every retail brokerage has to answer one question before it ever quotes a price: when a client clicks buy, what happens to that order? The answer defines the broker's dealing model, and it shapes everything downstream — where revenue comes from, what risks sit on the balance sheet, and how the firm thinks about its relationship with the people who trade through it.
The debate is usually framed as A-Book vs B-Book, as if one were honest and the other suspect. That framing is misleading. Both are legitimate, widely used, and regulated. The real distinction is about who carries the market risk and how disciplined the broker is at managing it. This guide explains each model, the hybrid approach most serious brokers actually run, and what the choice means for traders.
What A-Book Means
An A-Book broker passes client orders through to the wider market. When a client opens a position, the broker places an offsetting trade with a liquidity provider — a bank, a prime broker, or an aggregated pool of venues — so the firm holds no net directional exposure to that client's position. The client's profit is the liquidity provider's loss, and vice versa; the broker sits in the middle as a conduit.
Because the broker is hedged, it is neutral to whether the client wins or loses. Its revenue does not depend on client outcomes. Instead, an A-Book broker earns from:
- Spread markup — adding a small margin to the raw price received from the liquidity provider.
- Commission — a fixed or volume-based fee per lot traded.
- Routing rebates — payments from venues for directing flow to them.
The appeal of pure A-Book is alignment. The broker makes money when clients trade, not when they lose, which removes the most-cited conflict of interest. The trade-off is thinner per-trade margins and dependence on volume. A-Book also exposes the broker to execution costs it cannot always pass on cleanly: slippage on the hedge, rejected fills during volatile windows, and liquidity-provider minimums that make very small accounts uneconomic to hedge one-for-one.
What B-Book Means
A B-Book broker internalises the order — it takes the other side of the client's trade itself rather than hedging it externally. The position lives on the broker's own book. If the client loses, the broker keeps the difference as revenue; if the client wins, the broker pays out from its own capital.
This is where the conflict-of-interest framing comes from, and it is worth stating plainly rather than dismissing. Under B-Book, the broker's revenue includes net client losses, so a poorly governed firm could in theory benefit from clients failing. That is the legitimate concern, and it is why B-Book attracts scrutiny.
But the framing misses why B-Book is standard practice. Internalising flow is not a trick; it is a capital and risk decision with real advantages when run properly:
- Better execution for the client. The broker can fill instantly at the quoted price without waiting on an external venue, often with tighter spreads and no rejections.
- Economic viability of small accounts. Most retail flow is too small to hedge externally at a sensible cost. Internalising it is frequently the only way to serve those clients at all.
- Natural netting. A book of clients includes longs and shorts on the same instrument that offset each other. The broker only carries — and only needs to hedge — the residual net exposure.
The key word is net. A reputable B-Book operation is not gambling against its clients position by position. It is running a portfolio, watching the aggregate exposure, and hedging the residual it cannot absorb. The discipline that separates a sound B-Book desk from a reckless one is risk management, not the model itself. That discipline — exposure monitoring, trader classification, and defence against toxic flow — is exactly what dealing-desk risk frameworks are built to enforce.
Hybrid Dealing and Smart Order Routing
In practice, very few brokers are purely one or the other. The dominant approach is hybrid: the broker classifies its order flow and routes some trades to the A-Book and keeps others on the B-Book, often dynamically and at the level of individual clients or even individual orders.
The logic is straightforward. Not all flow carries the same risk. Internalising consistently profitable or predatory flow is expensive; internalising the broad base of retail flow that nets out is profitable and improves execution. A hybrid dealing desk sorts the two and routes accordingly.
Routing decisions typically rest on a few signals:
- Trader profile. A client's historical edge, win rate, and strategy fingerprint inform whether their flow is worth internalising.
- Instrument and size. Liquid majors net easily and suit the B-Book; large or illiquid orders may be safer hedged.
- Aggregate book exposure. When net exposure on an instrument breaches a threshold, the desk routes incremental flow to the A-Book to cap risk.
- Market conditions. During high-volatility events, a desk may widen the share of flow it hedges externally.
This is where smart routing and bridge configuration on the trading platform does the heavy lifting, translating risk policy into per-order execution decisions in real time.
Trader Classification and Toxic Flow
The engine underneath every hybrid desk is trader classification — the process of profiling flow to decide how to handle it. The categories that matter most to a B-Book are the ones that lose money predictably and the ones that win money predictably; the broker happily internalises the former and routes the latter out.
Two patterns demand particular attention:
- Latency arbitrage. Traders, often automated, exploit the gap between the broker's quoted price and the true market price during fast moves. They are effectively picking off stale quotes. Against a B-Book, every one of those trades is a near-guaranteed loss for the broker.
- Deposit hunters and bonus abuse. Accounts engineered to extract value from promotions or pricing gaps rather than to trade in good faith.
This kind of flow is called toxic because it is consistently unprofitable for the counterparty taking the other side. A B-Book that fails to identify and reroute it bleeds capital. Defences include execution-speed checks, quote-freshness validation, configurable HFT filters, and routing rules that automatically push flagged accounts to the A-Book. Building those controls is core to dealing-desk risk management, and they are also a major reason naive "just B-Book everything" operations fail while disciplined ones thrive.
Comparing the Models
| Dimension | A-Book | B-Book | Hybrid |
|---|---|---|---|
| Order handling | Passed to liquidity provider | Internalised on broker's book | Routed by classification |
| Market risk to broker | None (hedged) | Carries net exposure | Managed residual |
| Revenue source | Spread markup, commission, rebates | Spread plus net client losses | Both, by flow type |
| Conflict of interest | Minimal | Present, must be governed | Managed via routing |
| Execution quality | Depends on LP fill | Often instant, no rejects | Best-of-both, by order |
| Small-account economics | Often uneconomic to hedge | Efficient | Efficient |
| Capital requirement | Lower | Higher (must cover payouts) | Moderate |
| Operational demand | Lower | High (risk discipline) | Highest (classification engine) |
No row in that table makes one model strictly superior. A-Book trades margin for neutrality and simplicity. B-Book trades risk and capital for richer margins and better execution. Hybrid captures most of the upside of each but demands the most sophisticated infrastructure to run safely.
What It Means for Traders
For a trader, the model behind the platform is worth understanding but rarely worth fixating on. Under A-Book, the broker is indifferent to your results, but you may see occasional requotes or slippage as orders pass to external liquidity. Under a well-run B-Book, fills are typically faster and spreads tighter, and the conflict of interest is bounded by the broker's own risk limits and regulatory obligations — most retail flow nets out, so the firm has no stake in any individual trader's loss.
The signal that actually matters is not the label but the conduct: transparent pricing, consistent execution, clear regulatory standing, and no pattern of worsening conditions precisely when an account becomes profitable. A consistently winning trader on a hybrid desk should expect their flow to be routed to the A-Book over time, which is a perfectly normal outcome, not a penalty.
The Choice Is a Risk Decision
A-Book, B-Book, and hybrid are not moral positions; they are risk-management postures. The question a broker is really answering is how much market risk it wants to carry, how much capital it can commit to backing that risk, and how good its tooling is at telling profitable internalisable flow from toxic flow that must be hedged. Get that classification and exposure discipline right and B-Book is a sound, client-friendly business; get it wrong and even a conservative book can be drained by latency arbitrage and unmanaged net exposure.
That is why the dealing model should be designed alongside the rest of the brokerage rather than bolted on. The platform routing, the risk engine, and the classification logic are one system — a point worth weighing when you plan how to start a brokerage and when you choose between MT5 white-label or an alternative to host it. PropHub builds these pieces — dealing-desk risk, routing, and the broker platform stack — to be configured as a single coherent policy rather than three disconnected switches. Whichever model a broker lands on, the model is only as good as the discipline behind it.