Execution / Brokers
Broker execution explained: A-book, B-book and what slippage really is
Two brokers can quote the same price and give you two different results. The reason is execution: who ends up on the other side of your trade, and what happens when the price moves. This is the part of broker selection that marketing pages describe vaguely and that a live account reveals immediately.
Why the same strategy fills differently at two brokers
A trading system does not send “buy one lot”. It sends a request into a specific venue, through a specific route, priced from a specific feed. Change the broker and you have changed all three, even if the chart looks identical. FX and spot metals make this worse than equities or futures because the market is over-the-counter: there is no single central exchange to anchor one official price, so the venue you can reach is your market.
That is why “spread” is a poor headline number. What actually matters is the price you get when the market is moving, which is exactly when the differences appear and exactly when a backtest stops being representative.
Quiet market
Almost any broker fills well. Spreads are tight, requotes are rare and execution differences are invisible.
Fast market
Here the models separate. The last price is gone, the order is either slipped, rejected or filled at the new level.
Your strategy
If it holds for days, it will never notice. If it holds for seconds, the fast-market behaviour is the whole result.
A-book, B-book and the hybrid in between
These three words describe where your order ends up, and therefore whose interests are aligned with yours.
A-book
Your order is passed to a liquidity provider. The broker earns from the spread markup or a commission; it does not care whether you win. Execution quality is the product.
B-book
The broker takes the other side. Your loss is its revenue. Pricing, requotes and rejections become commercial decisions as much as technical ones.
Hybrid
Most retail brokers run both and route by client profile. The same account can be internalised on some trades and passed upstream on others.
B-book is not automatically fraudulent — many firms hedge and many pay winners without complaint. But it is a structural conflict of interest, and on a strategy that relies on capturing small, frequent moves, the conflict lands squarely on the edge.
STP, ECN, DMA — what the labels mean
The vocabulary is used loosely in marketing, so it helps to separate the three ideas cleanly. STP (straight-through processing) means your order is passed upstream rather than internalised. ECN means the venue allows your order to interact with orders from other participants. DMA, or direct market access, means your order reaches the venue’s market directly and you pay a commission rather than an embedded markup — see our DMA page for the mechanics.
What to ask
“Is this order internalised or routed?” and “what is the commission?” A broker that answers both directly is telling you something useful.
What to ignore
Average spread tables, “no slippage” claims and any execution statistic that is not published per order.
What to verify
The order type. A venue that supports fill-or-kill over a FIX API lets you refuse a bad fill; a venue that always fills cannot.
Slippage, requotes and rejections
Slippage is the difference between the price you expected and the price you received. It is not a malfunction; it is the market moving during your round trip. What varies is who absorbs it and whether you can decline it.
Slippage
Unavoidable in a fast market, but measurable. On a market-order platform it is invisible and compounding; on a route with fill-or-kill it is a decision.
Requotes
A requote is the broker declining your price and offering a worse one. It protects the broker rather than you, and it is a strong signal about the model.
Rejections
A rejected order is honest: the venue could not fill at your price. A platform that never rejects is a platform that always fills, which is not the same thing.
The cost number that actually matters
Cost per round trip is spread plus commission plus slippage, expressed in pips or in currency, on your typical trade. For a short-horizon system this single number decides whether the strategy is viable — not the average spread on the marketing page, which is measured when nothing is happening.
It is worth doing the arithmetic before the first live trade: if the average move you capture is three pips and the round trip costs two and a half, the system needs a very high win rate to survive, and any degradation in execution removes it entirely. This is the calculation XTRSK does on the quantitative trading page, and it is why execution is treated as part of the model rather than as plumbing.
Visible
Commission. A stated number per lot, identical on every trade.
Variable
Spread. Widen your sample to include news and rollover, or the number is fiction.
Hidden
Slippage. Only measurable if your route reports the price you asked for and the price you got.
How to evaluate a broker’s execution
A short checklist beats any review site, because it measures your strategy rather than someone else’s.
1. Route disclosure
Ask whether your order is internalised. Get it in writing, and check whether the answer changes by order size.
2. Order types
Confirm fill-or-kill and immediate-or-cancel are supported natively, not simulated by the platform.
3. Protocol
Is there a real FIX API, or a bridge pretending to be one? See what a FIX API connector does.
4. Fill log
Run a small live sample and log requested price versus filled price. Ten trades in a fast session tell you more than a year of backtests.
5. Rejection behaviour
Count the rejections. Zero rejections in a fast market is not a good sign; it means the venue is filling you at whatever price it likes.
6. Governance
Check how the firm describes its conduct obligations — the FX Global Code is the industry reference.
How XTRSK routes
XTRSK reviews the venue, the broker and the server location against the strategy before an account trades, and the systems run over FIX 4.4 with direct market access rather than through a bridged retail platform. PULSE-X takes fill-or-kill orders on a controlled latency budget; NIC-X takes displaced prints or nothing. The live account is published so the execution result can be checked rather than taken on trust.
Broker execution questions
Is a B-book broker bad?
It is a different business model with a built-in conflict of interest. What matters is whether the conflict is disclosed and whether winners are handled fairly — and, for short-horizon systems, whether the fills are good enough.
Why do I get requoted?
Because the price you asked for no longer exists. A venue that requotes is protecting itself; a venue with fill-or-kill over a FIX API lets the order die instead, which is usually better for the strategy.
Can slippage be eliminated?
Not eliminated — declined. With a fill-or-kill order you either get the price or you do not trade. That converts an unbounded cost into a binary outcome.
Is a lower spread always better?
No. A tight advertised spread on a B-book or bridged model can cost more in slippage and requotes than a slightly wider DMA spread with a stated commission.
How many trades do I need to judge execution?
A few dozen in a genuinely active session. The point is to sample the fast market, not to collect a comfortable average from quiet hours.
Further reading
The industry conduct standard for FX execution is the FX Global Code, published by the Global Foreign Exchange Committee. For the protocol used to route orders to a venue, see the FIX Trading Community standards.
This page explains execution mechanics; it is not a recommendation of any broker or venue, and it is not investment advice. Leveraged FX and metals trading can result in substantial losses.
Chat with XTRSK
Chat ready
Start a chat and the XTRSK team will be notified immediately.