The UK’s CFD loss warnings span 51% to 78%—and the gap is not a broker ranking

A prospective UK CFD customer can encounter loss warnings that appear to rank providers from substantially better to substantially worse. In an FXBrokers.ai snapshot checked on 9 November 2026, the selected provider pages displayed 51% at eToro, 69% at Capital.com, 69% at CMC Markets, 70% at IG, 73.6% at Pepperstone, 76% at Plus500, 76.6% at OANDA and 78% at XTB. The range was 27 percentage points.[2][3][4][5][6][7][8][9]

The arithmetic is stark. Across the eight warnings, the unweighted mean was 70.4% and the median 71.8%. Seven of the eight providers clustered between 69% and 78%; eToro’s 51% was the clear low observation. Excluding it, the seven-provider average was 73.2%. Those calculations are original, but they do not justify the conclusion that customers are 27 percentage points more likely to succeed at one provider than another.[2][3][4][5][6][7][8][9]

The Financial Conduct Authority’s methodology explains why. Each percentage is calculated from the provider’s own active retail accounts over a trailing 12-month period and refreshed every three months. The population, products, customer behaviour and market window can differ. The warning measures whether an account ended the calculation below zero after specified trading results and costs. It does not control for who opened the account, what they traded or how much they risked.[1]

What the FCA percentage actually counts

Under COBS 22.5, a retail account is classified as losing when the sum of realised and unrealised net profit on the covered speculative investments during the 12-month calculation period is below zero. The calculation includes costs, fees, commissions and other charges. Accounts without an open covered position during the period are excluded, as are profits and losses from other investments, deposits and withdrawals.[1]

The unit is an account, not an individual person. A customer may hold more than one account, and a provider’s population can change as new accounts arrive or inactive accounts fall outside the calculation. The measure also gives the same binary classification to an account down £1 and an account down £100,000. It reports frequency, not severity.[1]

An account at exactly zero is not below zero and therefore is not classified as losing under the rule. That feature can matter when accounts trade very little or when gains and costs offset. The warning still provides a clear high-level message—that most active retail CFD accounts lose—but the percentage is not the average return, median loss or probability attached to a single future trade.[1]

The eight-provider snapshot

At the low end, the eToro page displayed 51%. Capital.com and CMC each displayed 69%, and IG displayed 70%. These four values sat at or below the eight-provider mean. Pepperstone displayed 73.6%, Plus500 76%, OANDA 76.6% and XTB 78%. The percentages were transcribed from the cited provider pages rather than from comparison sites or affiliate tables.[2][3][4][5][6][7][8][9]

The decimal values at Pepperstone and OANDA suggest providers may publish the exact calculated result, while several others display whole numbers. Rounding alone cannot explain the 27-point range. It can, however, produce small apparent differences between providers whose underlying results are close. A displayed 69% and 70% should not be treated as a proven one-point performance advantage without knowing the unrounded values and calculation dates.[3][4][5][6][8]

The warning is provider-specific, but a brand can contain multiple legal entities and websites. A percentage on a UK page should not be applied to customers of a Cyprus, Australian or offshore entity. Product availability and client populations differ by jurisdiction. The relevant comparison begins with the exact contracting entity and warning, not the global brand name.[1][2][3][4][5][6][7][8][9]

Why customer mix can move the result

A provider attracting experienced, heavily capitalised traders may report a different loss frequency from one acquiring many first-time customers. That does not prove the platform caused the difference. Account size, frequency, leverage, holding period and instrument selection can all alter outcomes. Marketing channels can matter too: a campaign focused on short-term crypto or index speculation may bring a different population from one focused on hedging or lower-frequency trading.[1]

The proportion of dormant or lightly active accounts also affects the eligible population. Accounts with no open covered position in the period are excluded, but an account with limited activity can still enter the calculation. A small gain classifies it as non-losing; a small cost classifies it as losing. Providers with different activation and retention patterns can therefore produce different percentages even if execution quality is identical.[1]

Product mix is another confounder. Spread bets, rolling spot forex, equity CFDs, index CFDs, commodity CFDs and restricted options can exhibit different volatility, financing and customer behaviour. Some of the cited warnings cover spread betting as well as CFDs; others refer to CFDs. A provider’s headline percentage is the aggregate result of the products within the applicable warning, not a product-by-product comparison.[1][3][4][5][6]

Why market timing can move it

The calculation covers a rolling 12 months and is repeated quarterly. A volatile quarter enters the window as another leaves. If a concentrated group of accounts profited or lost during a major market move, the percentage can change even when the broker’s platform, pricing and customer base remain broadly stable. The warning is therefore a moving historical statistic, not a fixed property of the firm.[1]

The rolling window also means two providers’ displayed percentages may not have been calculated on the same day unless they updated simultaneously. Both may comply with a three-month update cycle while their windows differ by weeks. That timing difference can be significant during fast markets. A precise league table would require calculation dates as well as percentages, but those dates are not consistently prominent beside the warnings.[1][2][3][4][5][6][7][8][9]

Web publication adds another practical problem: provider sites contain many pages, cached versions and jurisdictional templates. A customer may encounter a figure on a legal page and another inherited footer elsewhere during an update cycle. FXBrokers.ai used the cited principal UK-oriented pages for this snapshot. The observation date is essential because these values are designed to change.[2][3][4][5][6][7][8][9]

What the percentage can tell a customer

The warning is powerful in one respect. Every selected provider showed that a majority of covered retail accounts lost money, and the highest figure approached four in five. That is a robust counterweight to advertising built around opportunity, speed or access. The warning should cause a customer to test affordability, leverage, costs and the possibility of rapid loss before opening a position.[2][3][4][5][6][7][8][9]

It can also prompt better questions. What are the spread and overnight financing costs for the intended instruments? How does the provider execute orders in fast markets? Which entity holds the account? What margin-closeout and negative-balance rules apply? Does the customer understand that an unrealised loss at the calculation date counts in the provider’s statistic? Those questions are more actionable than choosing the lowest percentage in isolation.[1]

The statistic is less useful for estimating the amount at risk. A 51% provider could have fewer losing accounts but larger average losses; a 78% provider could have many small losses and a smaller group of large gains. The public warning does not disclose either distribution. It also cannot show whether losing accounts were overleveraged, traded one event or accumulated financing costs over months.[1]

What it cannot prove about a broker

A lower loss percentage does not establish better execution, fairer pricing or superior education. Those may contribute, but so can customer selection and product mix. A higher percentage does not establish misconduct. The FCA designed the warning as a consumer-risk disclosure, not an enforcement score. Treating it as a quality ranking gives the number a causal meaning the methodology does not contain.[1]

The figure also does not show the broker’s own gain from customer outcomes. Some providers hedge extensively; others internalise portions of flow within controlled limits. Revenue can come from spreads, commissions, financing, interest and trading performance. An account’s loss includes costs and market movements, but the warning does not allocate that loss between the provider, external market counterparties and changes in the underlying price.[1]

Nor can the warning compare unleveraged investing with CFDs. Profits and losses from other investments are excluded even when held in the same broader customer relationship. A broker offering shares and ETFs can have a growing investment business while its CFD warning remains high. The percentage applies to the specified speculative products and eligible accounts.[1]

Sample size and movement are hidden

The warning does not disclose the number of eligible accounts behind the percentage. A one-point change among 500 accounts represents five accounts; among 500,000 it represents 5,000. Without the denominator, readers cannot assess statistical stability or determine whether a shift came from a small population. Large providers may have more stable percentages, but scale alone does not guarantee a representative mix.[1]

Nor does the public figure show transitions. A provider can report 70% losing in consecutive quarters even if many previously losing accounts recovered and a different group moved below zero. The unchanged headline would conceal substantial churn. Account-level transition rates, while more complex, would distinguish persistent loss from rotating short-term outcomes and help regulators understand whether customers repeatedly re-enter after losses.[1]

The percentage is auditable by regulators, not reproducible by the public

Firms must retain the underlying account records for five years, giving supervisors an evidential trail. Outsiders receive only the final percentage. They cannot verify the eligible-account count, calculation date, treatment of account migrations or the allocation of costs. That asymmetry is normal for confidential customer data, but it means comparison articles should present the number as a regulated disclosure—not an independently audited performance score.[1]

The regulatory opportunity

The FCA could make the disclosure more comparable by requiring the calculation date, exact legal entity and covered products beside the percentage. Publishing account counts would show whether a figure rests on thousands or hundreds of thousands of accounts. Distributional measures—median loss, median gain and average loss as a proportion of deposited funds—would answer questions the binary warning cannot.[1]

Such additions would need careful design. Providers warned during the original policy process that performance might be attributed incorrectly to the firm rather than the market or customer. More data could improve understanding but also produce misleading rankings if populations remain unmatched. The solution is not to remove the current warning; it is to describe its denominator and limitations more completely.[1]

The finding

FXBrokers.ai’s eight-provider snapshot found a 27-point spread, a 70.4% mean and a 71.8% median. Those are meaningful facts about the historical account populations reported on 9 November 2026. They are not controlled evidence that the lowest provider offers a 27-point advantage. The FCA rule classifies accounts, includes costs, uses each firm’s own rolling population and refreshes quarterly.[1][2][3][4][5][6][7][8][9]

The correct reading is simultaneously stark and restrained. Retail CFD losses were widespread at every selected provider. The exact percentages varied substantially. Customers should take the first message seriously and the second cautiously. The warning is an effective alarm, but a poor league table. A provider cannot be judged responsibly from the number until entity, products, dates, customer mix and loss severity are known.[1][2][3][4][5][6][7][8][9]

Report an error or submit an update