Every UK broker publishes the percentage of its retail clients who lose money. Reading those numbers as a league table is the most common mistake in retail trading research.
What does the average actually tell you?
Across 14 FCA-authorised UK CFD brokers, the April 2026 risk disclosures average out at 69.9%, with a median of 71.0%, on an analysis of the published figures by The Investors Centre. Roughly seven clients in ten lose money. That figure gets quoted constantly, and as a description of the retail leveraged trading population it is fair enough.
Before leaning on either number, though, it is worth knowing what they are made of. Each of those fourteen percentages was compiled by the firm publishing it, under its own reading of the reporting requirement, over a period that does not align neatly with anybody else’s, using a definition of an active account that varies from firm to firm. Averaging fourteen figures assembled on fourteen slightly different bases produces something closer to an approximate summary than a measurement, and treating 69.9% as precise to the decimal point is reading it harder than it can be read. That applies to whoever compiled it as much as to anyone quoting it afterwards.
As a decision-making input it is close to worthless. It tells you the activity is hard, which you knew. It does not tell you anything about any individual firm, and it certainly does not tell you anything about you. An average is a fact about a population, and you are not a population.
What does the spread tell you instead?
The interesting number in that sample is not the middle, it is the width. Individual firms ranged from 51% to 82%. Thirty-one percentage points separate the best-looking disclosure from the worst-looking one, among firms all authorised by the same regulator, all selling broadly the same product, all bound by the same leverage caps and the same negative balance protection. That is a very wide gap for a set of firms operating under identical rules. It means something. The question is what, and the obvious answer is the wrong one.
Why would two regulated brokers differ by 31 points?
Not, mostly, because one is better at looking after clients. The dominant driver is who walks through the door. A broker whose marketing pulls in first-time traders with small accounts and high leverage appetite will publish a worse number than one whose client base skews towards experienced traders running larger, less leveraged positions. Neither firm has done anything to the outcome. They have different customers.
Product mix does the rest. A firm offering mostly major currency pairs and index CFDs will look different from one pushing exotic pairs and single-stock CFDs, because the instruments themselves have different loss profiles. So does account minimum, so does the length of the reporting window, and so does whether the firm’s clients tend to hold overnight or close intraday.
| What moves the published number |
Direction |
Is it about broker quality? |
| Client base skews inexperienced |
pushes it up |
No, it is who they market to |
| High average leverage used |
pushes it up |
Partly, if the firm encourages it |
| Exotic and single-stock instruments |
pushes it up |
No, product mix |
| Higher account minimum |
pushes it down |
No, it selects wealthier clients |
| Clients hold intraday, not overnight |
pushes it down |
No, financing avoided |
| Genuinely better tools and education |
pushes it down |
Yes, though the disclosure cannot tell you how much of the gap it explains |
Six drivers of a published retail loss rate. Only one is unambiguously a statement about the broker rather than about its customers.
Is the lower number evidence of a better broker?
It is weak evidence at best, and a firm advertising its comparatively good loss rate is doing marketing, not disclosure. The disclosure exists because the regulator requires a risk warning, not because it was designed as a comparison metric. Repurposing it as one is a category error, and it is an error the industry is happy to let readers make when the number happens to flatter.
There is a sharper version of the problem. A firm could improve its published rate by raising its minimum deposit, which selects for wealthier clients, without changing anything about the quality of its service. The number would fall. Nothing meaningful would have improved. Any metric that can be gamed by changing your customer filter is not measuring what its readers think it measures.
What would you need for it to be a league table?
You would need the loss rates broken down by client experience, account size, leverage used, instrument and holding period, so that you could compare like with like. None of that is published, because none of it is required. What is required is a single percentage covering everybody, which is the least informative way to present the underlying data. So the honest position is that the 31-point spread tells you the firms have very different customer bases, and tells you almost nothing about which one would serve you better. Anyone presenting the range as a ranking is reading more into it than it can carry.
Does that make the disclosure useless?
No, and this is where people overcorrect. As a calibration device it is excellent. If you are about to open a leveraged account believing you will be in the winning minority, a number between 51% and 82% at every single authorised firm in the sample is a useful corrective. There is no broker where most clients win. That is the finding.
It is also useful over time. A firm whose published rate deteriorates sharply across several reporting periods is worth a second look, because something in its client mix or product push has changed. The level tells you little. The trend occasionally tells you something.
What should you compare instead?
The things that are actually about the broker rather than about its customers. What a round trip costs in spread and commission on the instrument you intend to trade. What overnight financing runs at in the direction you intend to hold. Whether withdrawals arrive when requested. What happens to your fill when the market moves quickly. Those are properties of the firm, they are comparable, and they are almost never published in a form that lets you compare them.
Which is why they have to be measured rather than looked up. Every one of those figures needs a funded account and a completed trade behind it before it exists, which is a slow and expensive way to fill in a comparison table and the reason very few tables are filled in that way. One of the few places it has been is a comparison of CFD trading platforms open to UK clients, built by a site which opens and funds live accounts with its own money to test UK trading platforms rather than compiling rankings from providers’ published fee schedules, and whose CFD comparison sets the FCA-regulated options against one another on those measured costs rather than on the risk warning at the top of the page.
What the risk warning on the page is actually for
Read it once, take one thing from it, and move on. That one thing is this: at every authorised firm in the sample, most clients lost money. There is no version of the disclosure anywhere in the market where the majority win. That is the whole of its useful content and it is genuinely useful, because a great many people open a leveraged account having quietly assumed the opposite and have never seen a number that contradicts them.
Everything else you would use to choose between two brokers sits somewhere else entirely and has to be measured rather than looked up: what a round trip costs on your instrument, what financing runs at in your direction, whether the withdrawal turns up when you ask.
So use the warning to set your expectations and the measured costs to pick the firm. If an article ranks brokers on that percentage, including one that funds its own testing, it is putting the number to a job it was never built to do.
…
