What share of retail clients lose money at each CFD broker?

Of the 5 brands in this table with both a UK figure and an EU figure, 4 report a higher share of losing accounts in the EU and 1 the same figure on both sites. The plausible drivers are boring: the two entities have different client mixes, different product mixes (UK sites often count spread bets alongside CFDs), and different twelve-month windows. What the pattern does confirm is the point our entity checker exists to make: which company onboards you changes the disclosures you get, the protections you have and, apparently, the odds of the clients around you.

The offshore and Australian end of the same brands is the other half of the story. On 14sites we read, the risk warning carries no percentage at all. No offshore regulator requires the figure and Australia's rules do not mandate this sentence, so clients onboarded there trade without the one statistic their EU and UK counterparts are shown.

• • FXTM: 88%, “88% of retail investor accounts lose money when trading CFDs with this provider.” source. On an FAQ page, in three wording variants ("CFDs", "Online Forex and CFDs", "Online Forex/CFDs"), with no entity named beside it. The main fxtm.com pages show no percentage at all.
• • eToro: 51%, “51% of retail investor accounts lose money when trading CFDs with this provider.” source. On a cross-entity regulation page. The UK, EU and Australian paths redirect or return 404, so the figure cannot be tied to eToro (UK) Ltd, eToro (Europe) Ltd or eToro AUS Capital.

Not yet read (10 sites): Exness (all regional sites: HTTP 403 to every fetch); FP Markets (all regional sites: HTTP 403 to every fetch); XM (all regional sites: HTTP 403 to every fetch); Vantage Markets (vantagemarkets.com and vantagemarkets.co.uk: HTTP 403 to every fetch); IG (all regional sites: HTTP 403; figures held from indexed copies of 2026-08-25); AvaTrade (all regional sites: HTTP 403 to every fetch); HFM (all regional sites: HTTP 403 to every fetch); ThinkMarkets (EU (CySEC) site: Geo-gated to EU visitors); VT Markets (Australian site: HTTP 403); Eightcap (UK and Australian pages: Every UK/AU path served the EU page). These brokers block automated reads of their pages; their figures will be added from browser reads and each row will carry its read date.

• • It is a client statistic, not a broker score. A provider whose clients trade smaller, less leveraged positions will post a lower number regardless of its pricing or conduct. A lower figure is not a reason to choose a broker, and a higher one is not evidence of wrongdoing.
• • It counts accounts, not money. An account that lost a small sum and one that was wiped out both count once. The sentence says nothing about how much was lost.
• • Twelve months, refreshed quarterly. Regulators require the calculation over the previous twelve months, updated at least every three months, which is why two reads a quarter apart can differ.
• • Ranges are allowed in some regimes.“Between 73–89%” is a compliant disclosure for some entities. We show ranges as published and leave them out of the median.
• • Retail only. Clients who elect professional status are excluded from the count and also lose the protections the number sits beside.

• • Each record is the risk-warning sentence as displayed on the broker's own regional site on the read date, copied verbatim. We never round, paraphrase or average a broker's figure.
• • One record per legal entity and site. Where a brand shows the same figure on several sites of one entity, we hold one record and note the others.
• • Provenance is labelled on every row: read on site (our own first-hand read), partner-supplied (wording sent to us in writing by the broker's partner team), or indexed copy (the site blocks automated reads, so the figure came from a search-indexed copy of the page and is queued for a browser check).
• • The snapshot date is the most recent read in the table. The oldest reads, first in the re-check queue: IG (25 Aug 2026), IG (25 Aug 2026), IG (25 Aug 2026).
• • Commission weighting: 0%. Which brokers appear, and which figure we hold for them, never depends on a partner relationship. Where two figures exist for one brand we display the higher on partner surfaces, because understating a loss rate is the compliance risk.

• • The table covers brokers in our research database, not the whole market. Absence from it means we have not read that broker's figure yet, nothing more.
• • Figures are not strictly comparable across entities: product mix, client mix and calculation windows differ, and some regimes permit ranges.
• • A figure read from an indexed copy or supplied by a partner may lag the broker's current display; those rows are labelled and will be replaced by first-hand reads.
• • Informational only, not financial advice. The percentage that applies to you is the one on the site of the entity that opens your account; read it there before depositing.

Clarytrade, a broker-research database, compiled the retail-loss disclosures of 11 CFD brokers (22entity-level records) from the brokers' own sites. Point figures range from 65% to 81% of retail accounts losing money (median 72.95%); where a brand publishes both, the EU figure exceeded the UK figure in 4 of 5 cases. Snapshot: 2026-09-01 (Q3 2026). Method: clarytrade.com/methodology. Verification queries: hello@clarytrade.com.

Please link to clarytrade.com/research/cfd-loss-rates when citing, and quote the snapshot date: the figures are refreshed every quarter.