Visit FxProAbout this desk
Broker Lab is a measurement desk, not a review site. We open ordinary accounts at retail brokers, run their own platforms on hardware we control, sample what those platforms send us, and publish the aggregate with the method attached. There is no star rating anywhere on this site and there never will be: a star is an opinion wearing the clothes of a measurement.
Last reviewed 13 October 2026
What we actually do
The work is unglamorous and it is most of the project. An account is opened at a broker in the usual way, on the account type a retail client would be offered. The broker's own platform is run on a machine that does not sleep. Bid and ask arrive from that platform and are stored raw, with the timestamp the server put on them. Every series is then audited for continuity before a single figure from it is published: a price that never visits the values next to its usual one is a fixture inside a demo server rather than a market, and nothing is averaged out of it.
Out of that store come the figures on this site: average spread through the liquid session, the spread hour by hour, the rollover band, and the annual cost of ownership those imply for a stated trading profile. Each one prints the window it was measured in and the method next to it, and each one carries the class of its source — our measurement, a third party, or the broker's own published terms. Those three never share a ranking.
The dataset behind the tables currently covers 12.10 19:08–13.10 17:13 UTC on EUR/USD for 3 brokers. Where a broker's series fails our feed audit, that broker is not ranked at all, however much work went into collecting it.
Who “we” is
A desk, not a cast of characters. You will not find author portraits or invented biographies on this site, because the thing that should carry the weight here is the method: the rig, the audit, the quarantine rules, the published window. A name attached to a number adds nothing a reader can verify, and a fabricated one is simply a lie in a place where our entire argument is that we do not tell them.
What can be checked instead is everything about how a figure was produced. The methodology page is written at a length most sites would consider excessive, including the three unit traps that produced plausible wrong numbers, the clock bug that stamped a Friday price with a Saturday time, and the demo feed we found that turned out to be a simulator rather than a market. Those are in there because the places a method can go wrong are the only part of it worth reading.
How the desk is paid
One broker, FxPro, pays a commission when a reader opens an account through our link. That is the whole commercial model, there is exactly one such broker, and it is labelled as sponsored wherever it appears. It is not in the comparison tables: we sampled its two demo platforms the same way as everyone else, the price they quote never moved, and a fixture is not a measurement. So it gets a card that says so rather than a row it has not earned.
No other broker on this site is linked, paid for, or aware of us. Nobody can buy a position, a removal, or a kinder figure. If that ever changes, it will be written on this page and in the disclosure before it appears anywhere else on the site.
- Read the method firstevery trap we fell into is written down
- Recompute the headlinethe trading profile is printed on the page that uses it
- Ask for the raw rowsthe data page says what we hand over and how
- Write to us about a numbercorrections are published, not quietly patched
Nothing on this site is financial advice, and nothing on it is a recommendation to trade. Trading leveraged products carries risk and you can lose money. Figures describe what we measured in a stated past window on our own accounts; your own broker, account type and conditions will differ.