CFD and leverage loss statistics: what 68%, 80% and 82% actually measure
Published retail CFD loss rates cluster between 68% and 82%, and they are the most cited — and most misread — numbers in retail trading. ASIC found 68% of Australian retail CFD investors lost money in FY2024. The FCA states approximately 80% of customers lose money on CFDs, with a sample finding 82%. Firm disclosures include 70% and 69%. What these figures measure is the share of accounts or investors ending a period with a net loss on leveraged contracts. What they do not measure is AI tool performance, unleveraged investing, or your personal probability of profit.
Published 27 August 2026 · Updated 28 August 2026 · AI Trading Book Editorial · Reading time about 14 minutes
- Five published figures, one range: 68% (ASIC), ~80% and 82% (FCA), 70% (IG), 69% (IBKR UK).
- They are not comparable — different periods, populations, product mixes and firm mixes.
- They say nothing about AI. No published series separates AI-assisted from discretionary trading.
- Leverage limits changed the magnitude, not the direction. Net losses fell 91% after ASIC's intervention; 68% still lost money.
- A frequency is not a personal probability, and the percentage tells you nothing about how much was lost.
The five published figures
Each of these comes from a regulator publication or a mandatory firm disclosure. None is an estimate, a survey or a vendor claim, which is why they anchor almost every honest discussion of retail trading outcomes.
| Source | Figure | Unit measured | Scope |
|---|---|---|---|
| ASIC REP 828, 20 January 2026 | 68% | Retail CFD investors | Australian market, FY2024 |
| FCA, general statement | approximately 80% | Customers | UK, CFD investing generally |
| FCA, CP16/40 sample | 82% | Client accounts | The sampled dataset in that consultation |
| IG, COBS 22.5 disclosure | 70% | Retail investor accounts | That firm, its disclosure period |
| Interactive Brokers UK, disclosure | 69% | Retail investor accounts | That firm, its disclosure period |
| New Zealand | Data not found | — | No equivalent published series identified |
What the figures precisely measure
Each number is a count of accounts or investors that ended a defined period with a net negative result on contracts for difference, divided by the total in that population. Three components carry all the ambiguity: which accounts are in the denominator, over what window, and on which products.
The UK firm disclosures exist because a rule requires them. Under PS19/18 and COBS 22.5, a firm offering CFDs to retail clients must display a standardised risk warning stating its own percentage of loss-making retail accounts. That is why the number appears on brokers' own websites: it is a mandated disclosure, not a marketing choice, and a firm cannot decline to publish it because it looks bad.
Why the figures are not comparable to each other
Presenting 68%, 80% and 82% in one table is useful for showing a range. It is not useful for ranking jurisdictions or firms, because at least four things differ between them.
- Unit. ASIC's figure counts investors; the UK firm disclosures count accounts. One person may hold several accounts.
- Period. A financial year is not a rolling twelve-month window, and market conditions differ between any two periods.
- Population. A market-wide regulatory measure covers every reporting firm; a firm disclosure covers one client base, shaped by that firm's pricing, minimum deposit and marketing.
- Product mix. A client base concentrated in major FX behaves differently from one concentrated in crypto CFDs, and the applicable leverage caps differ by a factor of fifteen.
The firm-level spread illustrates the point. IG at 70% and Interactive Brokers UK at 69% are close to each other and well below the FCA's approximately 80%. That does not establish that those firms produce better outcomes; it may reflect client mix, account minimums, product range or period. Reading a firm's disclosure as a performance league table is exactly the error the disclosure was not designed to support.
These numbers say nothing about AI
This is the most consequential misuse, and it runs in both directions. Vendors cite the loss rates to argue that human discretionary trading fails and their tool is the remedy. Critics cite them as evidence that AI trading does not work. Neither inference is available, because no published series separates method.
A retail CFD loss statistic counts every account at the reporting firms — people trading on charts, on news, on tips, on automated systems and on impulse — in one number. Extracting an AI-specific conclusion from it would require the regulator to have tagged accounts by trading method, and none does. Our evidence page covers what can actually be said about AI performance, which rests on fund index data and peer-reviewed work rather than on these figures.
What the numbers do establish is more useful anyway: in the retail leveraged market, the dominant driver of documented harm is the product and the leverage, not the sophistication of the decision process. That holds whoever, or whatever, is making the decisions.
What happened when leverage was capped
Australia ran something close to a natural experiment. ASIC's product intervention order took effect on 29 March 2021, capping retail CFD leverage, requiring 50% margin close-out and mandating negative balance protection. REP 724 measured what followed: aggregate retail client net losses fell by 91%, and the number of loss-making retail accounts fell by 51% per quarter.
A 91% reduction in aggregate net loss is among the largest documented effects of a retail conduct intervention anywhere, and it isolates the mechanism cleanly. Nothing changed about clients' skill, information or tooling between the two periods. What changed was the maximum size of the position they could take against a given deposit.
Both facts belong in any honest summary. The intervention removed a very large amount of harm. It did not make the activity profitable for most participants: three years later, REP 828 found 68% of retail CFD investors still losing money in FY2024. Capping leverage changed how much people lost, not whether they lost.
Why leverage produces these outcomes
The arithmetic is not subtle. At 30:1 leverage, a 3.3% adverse move eliminates the deposit backing a position. At 20:1 it takes 5%, and at 2:1 it takes 50%. Ordinary daily volatility in a major currency pair is a fraction of a per cent, so at high leverage the noise of a normal session is already a material share of the account.
Three mechanisms compound it. Costs are charged on the position, not the deposit — spread, commission and overnight financing all scale with the leveraged size, so a strategy that is marginally profitable before costs can be reliably unprofitable after them. Margin close-out realises losses at a moment chosen by the market rather than by the client. And the payoff distribution is asymmetric in a way that defeats intuition: a 50% loss requires a 100% gain to recover, so a sequence of moderate losses and moderate gains drifts downward even when the win rate looks respectable.
That last point connects directly to prediction accuracy. A model that is right 58% of the time sounds like an edge, and on our accuracy page we work through why it usually is not once costs, turnover and payoff asymmetry are applied — with leverage, the same win rate degrades faster.
The leverage caps in force
| Requirement | Australia (ASIC order) | United Kingdom (PS19/18) |
|---|---|---|
| Major currency pairs | 30:1 | Comparable limits in force |
| Minor pairs, gold, major equity indices | 20:1 | Comparable limits in force |
| Other commodities and minor indices | 10:1 | Comparable limits in force |
| Crypto-asset CFDs | 2:1 | Comparable limits in force |
| Margin close-out | At 50% of total initial margin | Required |
| Negative balance protection | Required | Required |
| Binary options for retail clients | Prohibited from 3 May 2021 | Prohibited |
| Mandatory loss-rate disclosure | — | Required under COBS 22.5 |
| In force | 29 March 2021 to 23 May 2027 | From 1 August 2019 |
The two protections that do the most work are the least discussed. Margin close-out caps how far a losing position runs before the provider intervenes — though it does not guarantee the exit price when a market gaps. Negative balance protection means a retail client cannot end up owing the provider more than they deposited, which before such rules was a documented outcome in violent moves. Neither protection applies to wholesale or professional clients, which is the substance of the warning on our Australian regulation page about accepting wholesale classification.
Five common misreadings
| Common reading | What the data actually supports |
|---|---|
| "I have a 32% chance of making money" | A historical population frequency is not a personal probability. Your outcome depends on product, leverage, costs, position sizing and holding period |
| "Australia is safer than the UK for retail traders" | The measures are constructed differently. The gap between 68% and 80% is not established as a difference in outcomes |
| "This broker is better — its disclosure shows 69%" | Firm disclosures reflect client mix, product range and period as much as anything the firm does |
| "Most traders lose, so AI tools must help" | Nothing here separates method. No AI-specific conclusion is available from these series |
| "Around 70–80% of investors lose money" | These figures cover leveraged CFDs, not investing. Unleveraged share ownership is a different product with different outcomes |
The survivorship problem inside the disclosure
One structural issue deserves naming because it is rarely stated on the disclosures themselves. A rolling twelve-month firm measure counts accounts active during that window. Clients who lost their deposit and stopped trading in an earlier period may not appear in the denominator at all.
The direction of that bias is not neutral. Excluding people who already lost and left tends to understate the failure rate of the activity over a longer horizon, because the population being measured is disproportionately made up of those still willing and able to trade. A figure of 68% or 70% for one year is therefore likely to be a floor rather than a ceiling for the lifetime experience of a cohort — though we found no published study quantifying that gap.
What we could not establish
- Any loss statistic split by trading method. No regulator or firm publishes a breakdown separating automated, algorithmic or AI-assisted accounts from discretionary ones. Data not found.
- A New Zealand equivalent. No comparable published series was identified. Data not found.
- A multi-year cohort study following the same retail clients. The published figures are period snapshots, not lifetime outcomes. Data not found.
- The average size of gain versus loss behind the headline percentages. The proportion is published; the distribution generally is not. Data not found.
Key takeaways
- The honest summary is a range, not a number: published retail CFD loss rates run from 68% to 82% across two jurisdictions and several measurement methods.
- Do not average them and do not rank with them. Unit, period, population and product mix all differ.
- They cannot support any claim about AI trading, for or against, because method is not separated anywhere in the data.
- Leverage is the demonstrated lever. Capping it cut aggregate net losses by 91% without changing anything about how clients traded.
- Reduction is not elimination. After the caps, 68% of Australian retail CFD investors still lost money in FY2024.
- A head count hides magnitude. The percentage says how many lost, never how much — and the missing distribution is where the real risk sits.
Frequently asked questions
What percentage of CFD traders lose money?
Published figures cluster between 68% and 82%. ASIC REP 828 found 68% of Australian retail CFD investors lost money in FY2024. The FCA states approximately 80% of customers lose money when investing in CFDs, with its CP16/40 sample finding 82%. Individual UK firm disclosures include IG at 70% and Interactive Brokers UK at 69%.
Are these loss statistics comparable to each other?
No. They differ in period, population, product mix and firm mix. The ASIC figure counts investors over a financial year; UK firm disclosures count accounts over a rolling twelve months at one firm. Placing them in one table shows the range, not a ranking of countries or firms.
Do these numbers measure AI trading performance?
No. They measure outcomes on leveraged CFDs across all retail clients at the reporting firms, whatever method those clients used. No published series separates AI-assisted from discretionary trading, so no conclusion about AI tool performance can be drawn from them.
Why do firms display a loss percentage on their websites?
Because a rule requires it. In the United Kingdom, COBS 22.5 under PS19/18 requires firms offering CFDs to retail clients to display a standardised risk warning including the firm's own percentage of loss-making retail accounts. The number is a regulatory disclosure, not marketing.
Why is the Australian figure lower than the UK figure?
The measures are constructed differently, so the gap is not necessarily a difference in outcomes. ASIC counts investors over a financial year across the market; the FCA figures derive from a different sample and period. Product mix, client mix and measurement window all differ.
Did ASIC's leverage caps reduce losses?
Substantially, on ASIC's own measurement. REP 724 reported that after the product intervention order took effect, aggregate retail client net losses fell by 91% and the number of loss-making retail accounts fell by 51% per quarter. Losses were reduced, not eliminated: 68% still lost money in FY2024.
What leverage limits apply to retail CFD clients?
In Australia, ASIC's order caps leverage at 30:1 for major currency pairs, 20:1 for minor pairs, gold and major equity indices, 10:1 for other commodities and 2:1 for crypto-asset CFDs, with 50% margin close-out and negative balance protection. The order runs to 23 May 2027. The UK has comparable limits under PS19/18.
What is margin close-out?
A rule requiring the provider to close a retail client's open positions when account equity falls to a set proportion of the initial margin required — 50% under ASIC's order. It caps how far a losing position can run before intervention, but it does not prevent a loss and does not guarantee the exit price in a gapping market.
What is negative balance protection?
A requirement that a retail client cannot lose more than the money in their trading account. Without it, a sharp adverse move can leave a client owing the provider more than they deposited. It is mandatory for retail CFD clients in Australia and the UK, and it does not apply to wholesale or professional clients.
Does a 68% loss rate mean I have a 32% chance of making money?
No. It is a historical population frequency, not a personal probability, and it says nothing about how much was won or lost. A cohort where 32% of accounts profit slightly and 68% lose heavily produces a large aggregate loss, which is broadly what the intervention data suggests.
Do the loss statistics include people who stopped trading?
That depends on the measure, and it is rarely stated plainly. A rolling twelve-month firm disclosure counts accounts active in that window. Clients who lost money and left before the window may not appear, which biases a snapshot toward those still trading.
Are binary options covered by these figures?
No. Binary options are a separate product, banned for retail clients in Australia from 3 May 2021 under a distinct product intervention order. The statistics on this page cover contracts for difference only.
Is there an equivalent loss statistic for New Zealand?
Data not found. No New Zealand equivalent of ASIC REP 828 or the FCA disclosures was identified. The Australian and UK figures should not be assumed to transfer, since product availability, leverage limits and client mix differ.
Does trading unleveraged shares carry the same risk?
No, and conflating the two is a common error. These statistics measure leveraged CFDs. Buying shares outright involves no margin close-out, no financing cost and no possibility of losing more than the amount invested. The loss rates on this page do not describe unleveraged investing.
Compiled by AI Trading Book Editorial from ASIC reports and media releases, FCA statements and policy material, and mandatory firm disclosures made under COBS 22.5. Figures are presented side by side with their measurement basis rather than averaged, because their scopes differ. Items we could not trace to a primary source are marked "data not found". Published 27 August 2026; last updated 28 August 2026. Corrections are logged on the corrections page.
Sources
- ASIC — REP 828, 20 January 2026 — 68% of retail CFD investors lost money in FY2024.
- ASIC — REP 724 — aggregate retail client net losses down 91%; loss-making retail accounts down 51% per quarter following the product intervention order.
- ASIC — 20-254MR, product intervention order effective 29 March 2021: leverage caps of 30:1, 20:1, 10:1 and 2:1; 50% margin close-out; negative balance protection. 22-082MR extending the order to 23 May 2027.
- ASIC — binary options product intervention order, effective 3 May 2021.
- FCA — "approximately 80% of customers lose money when investing in CFDs"; CP16/40 account sample 82%.
- FCA — PS19/18, permanent retail CFD rules in force 1 August 2019; risk warning and loss-rate disclosure requirements at COBS 22.5.
- IG and Interactive Brokers UK — firm-level retail loss disclosures of 70% and 69% respectively, published under COBS 22.5.
Informational research only. Nothing on this page is personal financial, legal, tax or investment advice, or a recommendation to trade any instrument. Contracts for difference are leveraged products that carry a high risk of losing money rapidly. Published loss rates describe historical populations and do not indicate any individual's likely result.