Law · Publication obligations
CFD loss-account percentages: calculation, requirements and interpretation
Legal and compliance information is translated from German. Reference language notice.
What the mandatory percentage actually measures, how a provider determines it and why it does not replace a broker ranking.
On this page
A percentage of accounts, rather than an amount lostWhy regulators require a provider-specific warningWhat the calculation includesA complete small account exampleOpen positions, costs and periodInterpreting the percentagePresentation: Visible, comprehensible and currentFrom the trade database to the published numberNew providers and missing dataLoss Rate and Copy TradingSourcesA percentage of accounts, rather than an amount lost
The percentage in the CFD risk indicator counts accounts with a negative net result. It does not measure what percentage of the deposited money was lost, how many individual trades failed or what the average loss was. The unit is the account; a person may have multiple accounts.
If seven out of ten included accounts lose, the odds are 70%. Whether each of these seven accounts loses one euro or a thousand euros, this quote does not change. Nor does it reveal how much the other three accounts have earned. Two providers can have the same quote and very different loss levels.
Why regulators require a provider-specific warning
The original ESMA intervention of 2018 created a standardized retail CFD protection architecture. Today, the respective applicable national measures are decisive; in Germany, the BaFin General Ordinance of 23.07.2019 is the central reference point. A historical ESMA decision alone does not fully describe the current legal ground of the German offer. The warning shall link the leverage indication to the actual share of retail losing accounts with the relevant provider.
[LOSS-ACT][LOSS-ORIGIN]The UK uses its own binding handbook rules for the relevant speculative retail products. Product scope, warning variants and presentation differ in detail. The percentage is not a worldwide requirement for any FX, securities or crypto provider. In the case of a cross-border offer, first determine legal entity, instrument, customer status and country of distribution. Regulatory audit trail
[LOSS-UK]What the calculation includes
The basic method is a calculation every three months covering the preceding twelve months. It includes the relevant retail CFD accounts that had an open CFD during that period. An account is classified as losing if its relevant realised and unrealised net result is negative. Trading-related costs are included. Deposits, withdrawals and results from other investment types are excluded from the CFD result. Accounts with no open CFD at any point in the period are excluded.
[LOSS-ACT][LOSS-ORIGIN]For the technical explanation, we write: Loss ratio = number of included accounts with net result below zero / number of all included accounts × 100. An exactly balanced account remains in the denominator, but does not belong to the loss counter. In principle, an account with only one relevant CFD counts once as well as an account with many relevant trades.
It is not a quote per trading volume. Large accounts must not be weighted more because they generate more sales. An account contract concluded during the period must not disappear from the technically relevant population solely because of its closure. The exact account and legal entity accruals must be documented in accordance with the relevant rules.
INTERACTIVE CONTOLABOR
How does the mix of accounts change the percentage?
Fictitious, already correctly delineated retail CFD accounts in the twelve-month window. Each included account counts once.
The display uses two decimal places for explanation. It does not specify a binding rounding or ready-made regulatory warning template. Zero-included accounts do not result in a quote.
A complete small account example
| Account | Realised result before costs reported separately | Open result on valuation date | CFD costs reported separately | Net/classification |
|---|---|---|---|---|
| A | +500 € | −700 € | 100 € | -300 € · Loss |
| BB | +200 € | 0 € | 250 € | -50 € · Loss |
| C | −100 € | +300 € | 50 € | +150 € · no loss |
| DD | +100 € | 0 € | 100 € | 0 € · No loss |
| EE | No CFDs in the period | — | — | Not included |
The simplified example contains four enclosed accounts, two of which are negative: 2/4 = 50%. A shows why only closed profits are not enough. B shows how costs rotate the sign. D shows the difference between “no loss” and “profit”. E shows why a filled customer base is not a correct denominator.
An additional deposit of €5,000 on A does not cure the trade loss. It increases the account balance, but not the CFD success. Similarly, a payout is not a trading loss. In addition, if C were to make a profit on a stock, this profit would be irrelevant for the CFD size shown.
Open positions, costs and period
For open positions, the valuation and deadline are central. A provider requires traceable prices, contract sizes, currency conversion and allocation of financing entries. A mere screenshot of the balance value is not enough: balance can overlook open losses; equity can contain deposits and other instruments.
The historical justification of ESMA refers to the value of open positions at the end of the calculation period in the case of unrealised results. For borderline cases beyond the beginning of the window, one must not invent a separate, untested evaluation method. The productive algorithm must jointly implement the relevant injunction, current supervisory interpretation and data booking. Our account laboratory presupposes the correct delineation of results and does not replace this preliminary work.
[LOSS-ORIGIN]Costs shall not be duplicated. If the spread is already included in the result calculated from the execution prices, it may not be deducted flat-rate again. Separately booked commissions or financing must be consistently included. The other fees attributable to CFD trading are part of a documented assignment rule; it is not a free choice according to the desired quote. Understanding Spread, Markup and Fees
Interpreting the percentage
| Statement | Classification |
|---|---|
| “80% lose 80% of their money.” | Wrong: frequency of negative accounts and loss amount are different sizes. |
| “The remaining 20% are successful.” | Too far: Zero results may be included; period and cost remain important. |
| “My probability of losing is exactly 80%.” | Underivable: historical provider population, not a personal forecasting model. |
| “The 60 percent provider is safer than the 75 percent provider.” | Not only usable: customer mix, instruments, activity and time period differ. |
| “A low quote proves fair execution.” | Unprovable: Execution requires its own data and tests. |
| “A high quote proves manipulation.” | Unprovable: Risk, leverage, behavior and costs can explain losses. |
Our editorial audit question is therefore: Which company reports which accounts for which time window and which product universe? The comparison should have the same customer group and, if possible, comparable reference dates. Even then, it remains a rough result frequency. Improvement can occur through other customers or products without the execution becoming better.
Presentation: Visible, comprehensible and current
ESMA’s historical interpretation calls for eye-catching warnings in the language of communication and at least the predominant font size. She describes fixed, repeated warnings in the website flow as good practice; hidden FAQ texts, small fonts, troubled backgrounds, and only pop-ups appearing on the first visit as bad practice. The current national legal basis and its interpretation must also be examined.
[LOSS-PROMINENCE]In the relevant website/app case, UK COBS requires a static warning visible when scrolling at the top of the screen and on websites the integration of linked pages. Warning variants depend on the medium and product. Shortening is not a free advertising decision. The account records used shall be kept there for five years.
[LOSS-UK]For operation, this means as an editorial implementation recommendation: a released warning component per jurisdiction, product and language; the same valid percentage value in website, app, landing pages and released campaigns; readability even for small displays and zoom. A minimal line in the footer is not a reliable design concept. Affiliate content and embedded ads must not fall out of the sharing process.
From the trade database to the published number
| Step | Practical implementation recommendation | Evidence |
|---|---|---|
| 1 · Population | Delineate legal entity, retail status, accounts and product scope | Documented rules and exclusion grounds |
| 2 · Data | Consolidate trades, open positions, costs, valuations and cash flows | Date export and data origin |
| 3 · Result | Calculate per account; detect double deductions, cash flows and third-party products | Calculation of accounts with reconciliation to the trading/ancillary book |
| 4 · Control | Sampling, prior period comparison, borderline cases and independent clearance | Test report and version of the algorithm |
| 5 · Publication | Roll out valid value and appropriate warning form across all channels | Release, time and archival documents |
| 6 · Follow-up cycle | Terminate quarterly runs; recognize delayed or conflicting ads | Protocol of Amendment and Escalation |
This process architecture is our implementation recommendation, not an alleged wording of a technical supervisory regulation. CSV exports alone do not guarantee completeness: migrated servers, multiple white labels, account currency exchanges or closed accounts can create gaps. Particularly important is the reproducible calculation – not just the last percentage.
A statistic from the group should not appear unchecked as a provider quota of the specific contract company. Rebranding, server change or reopening also do not allow a freely selected restart of the history. Such cases require documented demarcation and, where appropriate, prudential clarification.
New providers and missing data
A missing relevant account balance means “quote unpredictable”, not “losing 0%”. The rules provide for special standard warnings. The historical EU architecture uses the predefined bandwidth of 74-89% in the relevant replacement note; UK works with a text warning variant for the relevant case. Do not copy a historic EU bandwidth into a UK warning field or promote a new appearance as a lossless provider.
[LOSS-ORIGIN][LOSS-UK]Missing data from an already active provider must be distinguished from this. A technically failed export does not automatically justify the rule for a provider without relevant trades. Data problem, professional predictability and permissible substitute warning are separate decisions.
Loss Rate and Copy Trading
A provider-wide CFD loss rate is not a current account of a particular signal provider. An attractive trader ranking does not cancel the product warning. Conversely, the provider-wide number does not allow for a direct conclusion as to how many users of a particular strategy lose. For this, the strategy population would have to be evaluated with its own, disclosed methodology.
The Crucial Bridge: Fees, leverage, and real-world follower execution can change a copied approach. Therefore, the account quota and a detailed strategy analysis complement each other. Social, Copy and Mirror Trading Holistically · MetaTrader Signals Practically Set Up
Sources & scope
Check the evidence.
- BaFin · CFD General Order, 23.07.2019 · Annex II
German binding framework of reference. Official search extracts checked; full call technically blocked. The original decree remains decisive.
Open the source ↗ - ESMA · Decision (EU) 2018/796 · Annex II
Historical, temporary EU intervention. Calculation system; no substitute for current national measures.
Open the source ↗ - FCA · COBS 22.5.6–22.5.9
Current handbook, tested 30.09.2026: calculation, warning forms, recording and presentation. UK special rules are not transferred to Germany unchecked.
Open the source ↗ - ESMA · Q&A 1991, 09.11.2018
Historical interpretation to highlight the warning; good and bad presentation practices.
Open the source ↗