Retail CFD loss rates by broker

Key findings

• Median 74%, mean 71.7%. Across 53 UK and EU broker entities (38 brands), the disclosed share of retail CFD accounts that lost money ranged from 54.4% (Swissquote Capital Markets Ltd, CySEC) to 84% (Indication Investments Ltd, trading as Libertex, CySEC).
• 35 of 53 entities disclosed 70% or more; 5 disclosed less than 60%.
• 26 of 53 were below 74%, the bottom of the 74–89% range of losing accounts that ESMA cited in 2018 from national regulators' studies, most of them made before leverage limits applied.
• FCA-regulated entities: median 69.5% (22). CySEC-regulated: median 75.5% (18). The groups serve different clients and products, so the gap describes the sample, not the regulators.
• The same brand can publish different figures for each entity. Plus500: 76% (UK), 80% (Cyprus), 82% (Estonia). Swissquote: 76.92% (UK), 54.4% (Cyprus). Pepperstone: 73.6%, 74.1% and 76.1%.
• No figure in Australia. None of the five Australian sites we checked (IG, CMC Markets, Pepperstone, Capital.com, OANDA) published a percentage: the ESMA and FCA rules don't apply there.

Distribution and regulators

Most entities cluster between 65% and 80%: 40 of 53.

Why brokers publish this figure

In 2018 the European Securities and Markets Authority restricted the sale of CFDs to retail clients across the EU (Decision (EU) 2018/796, applied from 1 August 2018): leverage limits, a margin close-out rule, negative balance protection, no trading incentives, and a standard risk warning. The warning must include “an up-to-date provider-specific loss percentage”, and the calculation “shall be performed every 3 months and cover the 12-month period preceding the date on which it is performed”. A provider that had no open retail CFD in the last 12 months uses a generic sentence instead: “Between 74-89 % of retail investor accounts lose money when trading CFDs.”

ESMA's measures were temporary and renewed every three months; between March 2019 and April 2020 national regulators replaced them with permanent rules of their own (ESMA, product intervention). In the UK the Financial Conduct Authority made the rules permanent in PS19/18 (1 July 2019, in force for CFDs from 1 August 2019), now COBS 22.5 of its Handbook. COBS 22.5.6R requires the percentage of retail client accounts that lost money, calculated “every three months” over “the 12-month period preceding the date of the calculation”. An account counts as losing if its realised and unrealised net profit on these products over the period is below zero, after all costs, fees and commissions; accounts with no open position in the period, deposits and withdrawals are left out. UK figures include spread bets where the firm offers them.

The figures are therefore updated quarterly by each firm. We check them again every quarter: the next check is in February 2027.

How we collected it

• The broker's own page only. We looked for the disclosure on about 60 broker websites serving retail clients in the UK, the EU and Australia, on their homepage or CFD page, on November 12, 2026 between 11:25 and 14:03 UTC, from a desktop browser in France. Third-party sites were used only to find where a disclosure sits, never as the source of a figure.
• One row per legal entity. Many brands run several regulated companies, each with its own figure; we recorded each one we could read separately, and identified the entity from the legal text on the same page.
• What a visitor sees. 48 figures come from the text of the rendered page, 3 (Trading 212) from the per-entity warning text in the page's code, and 2 (XM, RoboMarkets) from the page's HTML because it didn't finish loading in our browser.
• Conflicting figures are shown, not hidden. On 6 sites a second figure for the same entity also appeared that day (a hidden copy of the warning, an older string in the page's code, or another page of the site). We kept the figure of the visible risk warning and list the other in the table's notes and the CSV.
• Evidence kept. For every row we saved the URL, the exact sentence, the time it was read and the text that names the legal entity; the file is in our repository at scripts/studies/cfd-loss-rates/raw-2026-11-12.json.
• Statistics. Count, mean, median, lowest and highest across entities, unweighted: a firm with a few hundred clients counts as much as one with hundreds of thousands.

Limitations

• Self-reported, not audited. Each firm calculates its own figure under the regulator's method. The FCA's rule asks firms to keep the records behind it for five years; it doesn't ask for an independent audit.
• Different 12-month windows. Each firm recalculates every three months on its own schedule, so on the same day the figures cover 12-month periods that can end up to three months apart, and a firm may update its website later than its calculation.
• Client mix, not quality. A lower share doesn't mean a given client is more likely to profit with that firm. It depends on who the clients are, what they trade and for how long; professional clients are not counted.
• Losing, not how much. The figure counts accounts with a net loss over the period; it says nothing about the size of the losses or gains.
• A sample. These are the brokers whose figure we could read and attribute; sites can show other content in other countries, and some firms were left out (below).

Brokers left out

• eToro. Shows 55% once, in a footer that lists eToro (Europe) Ltd (CySEC), eToro (UK) Ltd (FCA), eToro AUS Capital (ASIC) and non-European entities, so the figure can't be attributed to one entity.
• AvaTrade. Shows 65% on the group site, whose footer lists AVA Trade EU Ltd (Central Bank of Ireland) alongside non-European entities, so the figure can't be attributed to one entity.
• ActivTrades (EU site). Shows 73%, but neither the page nor its regulation page named the legal entity when we read them.
• ThinkMarkets. The UK and EU pages redirected us to a French-language page run by a Seychelles entity; no UK or EU figure could be read.
• Exness (EU). exness.eu is a business-to-business page with no retail loss percentage.
• easyMarkets (EU). The page could not be loaded.
• Sites outside the UK and EU rules. IG's international site showed 80% and ActivTrades' non-European site 85%; both are outside the scope of this tracker, as are Capital.com's non-EU group entity and other firms that show ESMA's generic 74–89% sentence.
• Australia. IG, CMC Markets, Pepperstone, Capital.com, OANDA: no percentage on their Australian sites.

What this means if you build or buy an automated strategy

A bot trades the same CFDs as a person, through the same accounts, so these figures describe its starting point too. Three things decide most of the result, whoever places the orders:

• Costs compound with the number of trades. The spread, commission and overnight financing are part of the calculation above. Work out what they take from the strategy's average trade before anything else, for example with the trading fee calculator.
• Leverage multiplies losses as well as gains. The retail limits (30:1 on major currency pairs, down to 2:1) are a ceiling, not a target. Our test of 240 leveraged setups chasing +10% a day shows how quickly liquidations add up.
• Test out of sample before real money. A backtest that looks good on the data it was tuned on says little; keep a period apart and include realistic costs. The Backtest Lab does both, and our algorithmic trading statistics collect the research on how often backtested edges hold up.

A seller's backtest or a past track record is not a guarantee, and nothing on this page is investment advice.

Data and how to cite

cfd-retail-loss-rates-2026-q4.csv (53 rows: id, broker, legal entity, regulator, percentage, the exact sentence and its language, source URL, check date, other figures seen and notes). The data is free to reuse under the Creative Commons Attribution 4.0 licence: credit Edgoro and link to this page.

Edgoro, “Retail CFD loss rates by broker”, edgoro.com/research/cfd-retail-loss-rates, checked November 12, 2026. Data: edgoro.com/data/cfd-retail-loss-rates-2026-q4.csv (CC BY 4.0).

Spotted a figure that has changed, or a broker we missed? Write to us through the contact page. The tracker is checked every quarter; next check February 2027.

Trading involves substantial risk of loss and is not suitable for every investor. Edgoro is a software tool, not a broker or financial adviser: nothing it produces is investment advice, and past or backtested performance does not guarantee future results.