What can and cannot be verified
Execution quality is the hardest thing on a broker's page to check and the easiest to claim. A retail client sees their own fills and nothing else: no order book, no record of what the price was elsewhere at that microsecond, no way to distinguish a fair fill in a fast market from a poor one. Brokers know this, which is why 'lightning-fast execution' appears on almost every website and a latency figure appears on almost none.
So this ranking measures disclosure rather than performance. A broker that names Sucden Financial and INTL FCStone as liquidity providers has made a statement that can be checked and that would embarrass it if untrue. A broker that publishes order execution under 50 milliseconds has committed to a number. A broker that says it offers 'institutional-grade execution' has said nothing at all, and is ranked accordingly.
What the execution labels mean
- Dealing desk: the broker is your counterparty and manages its own risk against your position.
- STP: orders pass to liquidity providers without desk intervention, usually at a marked-up spread.
- ECN: orders meet other participants' orders on a shared feed, priced raw with a separate commission.
- NDD is an umbrella covering STP and ECN, and says nothing about which of the two you are getting.
The account-level trap
The most common misreading of an execution claim is treating it as firm-wide. It rarely is. ForexChief states that its STP/NDD model runs on MT4.Classic+ and MT4.DirectFX — naming the accounts, and implicitly excluding the others. ICM Capital separates ICM Direct, priced at 1.3 pips with no commission, from ICM Zero at 0 pips plus commission. Alpari's ECN routing applies to its ECN accounts. In each case the broker has been precise, and a reader who takes the headline model as applying to the default account will end up on a different one.
This matters more than it sounds, because the default account is where most new clients land. The account opening flow rarely presents the routing model as a choice; it presents deposit size and spread. If execution model is what you care about, it has to be selected deliberately at signup, and on several brokers it requires a higher minimum deposit than the default.
A no-dealing-desk claim is not a conflict-free guarantee
STP and ECN routing reduce the broker's direct interest in your losses; they do not eliminate every conflict. Brokers still earn from spread markup, commission, and in some cases payment for order flow to particular liquidity providers. Read the routing claim as information about structure, not as a promise of alignment.
Where execution actually degrades
Fills go bad in predictable places, and none of them are about the broker's servers. The first is the seconds around a scheduled release, when liquidity providers widen or withdraw quotes and a market order fills at whatever remains. The second is the daily rollover, when the book thins for a few minutes. The third is the Sunday open, where a weekend gap is filled against the first available price rather than your stop level. A guaranteed stop is the only instrument that removes this risk, and it is priced accordingly.
Order type does more for execution quality than broker choice in these moments. A limit order cannot fill worse than its price; a market order can. Traders who blame a broker for slippage on a market order placed thirty seconds before a rate decision are usually describing the market, not the firm — but a broker whose slippage is consistently one-directional, always against the client, is describing itself.
Common questions about execution
Can I test a broker's execution before funding a large account?
Partly. A demo account runs on the same feed but not the same routing, so it shows spread behaviour and not fill quality. A small live account traded through a few news releases is the only realistic test, and it is worth the cost before moving a serious balance.
Is ECN always better than STP?
Not for everyone. ECN gives raw pricing with a commission and suits larger or more frequent trades; STP with a marked-up spread is simpler and cheaper below roughly a third of a lot. The routing quality of a good STP broker is not worse — the pricing model is different.
What latency should a retail trader expect?
Anything under about 100 milliseconds end-to-end is beyond what discretionary trading can exploit. Latency matters for automated strategies and for scalping inside the spread; for a swing trader it is not a differentiator, whatever the marketing says.
Do brokers publish slippage statistics?
Very few, and those that do publish them aggregated across all clients and instruments, which hides the retail experience during volatility. Treat published slippage data as better than nothing and considerably less than proof.