Three acronyms dominate broker comparison pages. ECN, for electronic communication network. STP, for straight through processing. NDD, for no dealing desk. Traders treat them as a quality ranking, with ECN at the top, and choose accounts accordingly.
The terms do describe real architectures. What they do not do is guarantee that a given firm uses them, because none of the three is a defined legal category with a register behind it. A broker can print ECN on the pricing page on Monday and change nothing about its systems.
What the labels are supposed to describe
An electronic communication network matches orders from multiple participants against each other. Bids and offers from banks, funds and other clients sit in a shared pool, and the broker's role is to give you access and take a commission for it. Prices come from that pool, which is why raw spreads on such accounts can be extremely narrow and occasionally show a zero spread for an instant.
Straight through processing means your order is passed to an external liquidity provider without a human or a dealing desk deciding what to do with it. There is no shared pool; the broker takes your order and covers it with a counterparty, then hands you the result.
No dealing desk is the broadest of the three and means only that no person intervenes between the order and its execution. Automated internalisation is still no dealing desk. That is the gap most of the marketing lives in.
What actually happens to your order
Practically every retail broker of any size operates a hybrid. Some client flow is internalised, meaning the firm becomes the counterparty and manages the resulting exposure across its whole client base. Some is passed out to liquidity providers. Which flow goes where is decided by a routing system that classifies accounts and strategies by how they behave.
That sounds sinister and is mostly not. Offsetting one client's long against another's short is efficient and produces tighter pricing than sending both to an external venue and paying twice. What matters is whether the decision is made under a policy the firm can evidence, or informally in a way that puts the house on the other side of clients it expects to lose. The distinction is set out at length in A book and B book execution, and the routing machinery itself in order routing.
The question worth asking is not "does my broker ever take the other side". It is "does the firm's revenue depend on my losses, and can the regulator do anything about it if it does". Those are two different things, and only the second is answerable from public documents.
The cost arithmetic, done properly
Raw spread accounts and standard accounts charge the same thing in different places, and comparing them requires converting both to one number.
| Account type | Where the fee sits | How to compare it |
|---|---|---|
| Raw or ECN | Narrow spread plus a commission per lot each way | Convert the round-turn commission into pips and add it to the average spread |
| Standard | A markup built into the quoted spread | Use the average spread directly, measured at your trading hours |
| Both | Overnight financing on positions held past rollover | Check the swap table separately; it is unrelated to the execution label |
Do the measurement at the hours you actually trade. An average spread quoted as a headline number is usually a 24 hour mean that includes the quiet hours nobody trades, and the figure during the London or New York session can look quite different. Comparing at the wrong time of day is how traders talk themselves into an account that costs more. The composition of these charges is broken down in spreads, markups and commissions.
One consequence follows directly. A high frequency approach pays the fee many times a day, so a genuinely narrower total cost compounds into something significant. A trader taking three positions a week and holding them for days will find that overnight financing dwarfs the execution cost, and the whole ECN debate is close to irrelevant to their results.
What to check instead of the label
Start with the regulator. A firm authorised in a jurisdiction with enforceable best execution obligations has to publish an execution policy describing the venues it uses and the factors it weighs, and it can be examined against that document. A firm in a jurisdiction without such rules can describe itself however it likes with no consequence, and that difference matters far more than any acronym.
Then measure. Log the requested price and the fill price on every market order for a month and look at the distribution. Slippage that is roughly symmetrical, sometimes better and sometimes worse, is what an honest pass-through produces. Slippage that lands against you almost every time is a finding. Do the same for spreads around scheduled releases and compare against a second account at a different firm on the same days.
Ask direct questions in writing, too. Which liquidity providers supply pricing. Whether client flow is internalised and on what basis. Whether the firm applies last look, and what the rejection rate is. A firm with a straightforward answer will give you one. The article on liquidity providers explains what a credible answer looks like.
The honest summary of the three models
For most retail traders, the execution label is a weak predictor of anything they will experience. Regulation, published policy, measured slippage and the total cost at their own trading hours are strong predictors. A well-run firm that internalises flow under a supervised policy can give a better experience than an offshore firm that prints ECN on its homepage and routes nothing anywhere.
Where the label does earn attention is at the extremes. If you trade in seconds, need to see the pool your orders meet, and are large enough that fill quality moves your results, the architecture is a real consideration and worth interrogating in detail. If you take a handful of positions a week, the time is better spent on the things that actually decide the outcome: sizing, the cost of holding, and whether your money is segregated if the firm fails. Leveraged trading carries a high risk of loss whichever execution model sits behind the platform.
"Nobody has ever shown me an ECN certificate, because there is no such thing. Show me the execution policy and a month of my own fills, and I will tell you what kind of broker you have."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- ECN, STP and NDD are descriptions a broker chooses, not categories a regulator issues.
- Almost every retail broker runs a hybrid, internalising part of the flow and passing on the rest.
- Convert commission into pips and compare total cost at the hours you actually trade.
- Measured slippage symmetry over a month says more than any label on the pricing page.
Frequently Asked Questions
Is an ECN account always cheaper than a standard account?
Only if you do the arithmetic. A raw spread account charges a commission per lot on top of a very narrow spread, while a standard account hides its fee inside a wider spread. Add the round-turn commission expressed in pips to the average raw spread and compare that total against the standard account's average spread on the same pair at the same hour. For frequent traders the raw account usually wins; for occasional trades on exotic pairs it often does not.
Can a broker be ECN and still take the other side of my trade?
In most jurisdictions nothing stops a firm from using the label while internalising some client flow. There is no licence, certificate or register for the term, so it carries no enforceable meaning by itself. What is enforceable, where the firm is properly regulated, is the execution policy it publishes and the outcomes it delivers, which is why the policy document is worth more attention than the marketing page.
How do I test my broker's execution myself?
Record spreads at the same times every day for two weeks, including around scheduled news, and compare against another account. Log requested price against fill price on market orders and check whether slippage is symmetrical or falls consistently against you. Time your fills. None of this proves an execution model, but persistent one-sided slippage and spreads that widen only for you are the signals worth acting on.