ASIC’s 91% CFD-loss reduction still left A$458m of annual retail losses—and A$73m was fees

Australia’s contracts-for-difference intervention produced a striking result in its first six months: aggregate net client losses per quarter fell by 91%. Yet ASIC’s later sector review found that retail clients lost more than A$458 million in the 2023–24 financial year, including A$73 million in fees, and that 68% of retail CFD clients ended the year with a net loss. Both findings can be true. They measure different windows and answer different questions.[1][2]

The distinction matters because the 91% figure is easily read as proof that the underlying retail-loss problem was almost eliminated. It was not. The product intervention order constrained leverage, standardised margin close-outs, required negative-balance protection and prohibited certain inducements. Those measures sharply reduced the speed and scale of losses relative to the pre-order period. They did not convert leveraged over-the-counter speculation into a product on which most retail clients made money.[1]

The A$73 million fee component was about 15.9% of the A$458 million aggregate loss figure. That does not mean every client paid 15.9% of a personal loss, because the sector totals combine different account histories, gains, losses and charges. It does establish that costs were a material part of the aggregate outcome. A profitable position before financing and transaction costs can become unprofitable after them; ASIC specifically warns that leverage and financing fees can magnify or erode returns.[1][2]

A reduction is not the same thing as a low level

ASIC’s consultation says quarterly aggregate net client losses fell from approximately A$372 million to A$33 million in the initial comparison. The regulator also recorded an 87% reduction in margin close-outs and an 88% reduction in negative-balance occurrences. Those are consequential changes: a negative balance can leave a client owing more than the amount deposited, while forced close-outs can crystallise losses during a rapid move.[1]

But the starting level was extremely high. Cutting a large number by 91% can still leave a meaningful residual, and the later A$458 million figure covers a full financial year rather than one quarter. Multiplying A$33 million by four would produce A$132 million, but that is not a valid reconstruction of 2023–24. The client base, market volatility, trading intensity, product mix and fee burden changed. The early and later figures therefore should not be spliced into a trend line without the underlying quarterly series.[1][2]

Participation also changed. ASIC says the average number of retail clients trading per quarter in 2023–24 was about 76% below the average in the 12 months before the order commenced. That can reduce aggregate losses independently of any improvement in the outcome per trader. A complete evaluation needs at least four denominators: clients who traded, client-days or trades, capital deposited or exposed, and the distribution of gains and losses. ASIC’s public figures establish the direction of harm reduction, but they do not publish every denominator needed to isolate which mechanism contributed how much.[1][3]

The A$36 million margin-netting problem shows how implementation changed the result

The sector review found a more specific failure. Some issuers offered what ASIC called margin discounts when retail clients held opposing long and short CFD positions. The two positions could not produce a net trading profit while both remained open, but they could continue generating financing costs. ASIC concluded that more than half the reviewed sector had contravened the product intervention order in this area.[2][3]

Following ASIC’s intervention, 28 issuers refunded more than A$36 million to affected retail clients. Measured against the 52 licensed issuers in the regulator’s whole-of-industry review, that is about 53.8% of the reviewed population. The ratio should not be read as a finding that every one of those issuers committed an identical breach or caused an equal amount of harm; the refund total and issuer count aggregate different circumstances. It does show that a technical margin-calculation rule had direct cash consequences for customers.[1][2]

The proposed amendment would make the boundary explicit: initial margin must be calculated separately for each CFD, and long and short exposures under multiple CFDs cannot be netted to reduce the requirement. This is more than drafting housekeeping. If an issuer recognises the offset for margin purposes while continuing to charge both legs, it can make a locked position appear less capital-intensive without removing its carrying cost.[1]

Those percentages are calculated from ASIC’s count of 52 reviewed issuers. They measure remediation, not necessarily unique breaches: one issuer may appear in several rows, and a change can range from a narrow correction to a substantial redesign. The reporting row followed the identification of more than 70 million erroneous transaction reports, illustrating that weak data can obscure the regulator’s view even when the underlying client trade is already complete.[2][3]

The proposed order has a fixed statutory boundary

ASIC’s current product intervention order is due to expire on 23 May 2027. The regulator proposes, subject to ministerial approval, to extend it only until its statutory sunset on 1 April 2031. ASIC explicitly corrected the potential misunderstanding that an extension would restart a fresh sunsetting period. It is separately considering the legal route for maintaining protections after 2031.[1][4]

The amendments also respond to product evolution. ASIC proposes to cover derivatives with the same economic characteristics as CFDs whether settlement occurs in cash, by set-off or through the transfer of digital assets; to clarify the margin treatment of CFDs over options; to tighten the conditions and record-keeping around discounts; and to make clear that educational or research tools cannot be selectively used as inducements. It is also asking whether option CFDs that expose a retail client to losses beyond the premium should remain available at all.[1]

These proposals reveal where perimeter risk now sits. A rule written around a cash-settled contract can be bypassed in substance if the same leveraged payoff is packaged with another settlement method. A prohibition on inducements can be weakened if a benefit is described as education but made contingent on trading behaviour. And a margin rule can be undermined by portfolio presentation even when each contract remains separately costly.[1]

What the record proves—and what remains unresolved

The public record proves that the intervention substantially reduced several early harm measures, that most retail clients in the later annual sample still lost money, and that compliance failures led to refunds on a material scale. It does not prove that the order caused every change in participation or losses. Market conditions, issuer exits, client selection and changes in trading frequency are plausible alternative contributors.[1][2][3]

A stronger causal assessment would publish a consistent quarterly panel showing active clients, deposits, notional turnover, financing charges, gross trading P&L before fees, net P&L, close-outs and negative balances. It would separate new from experienced clients and identify how outcomes differ by leverage band and underlying asset. Evidence that loss rates and loss intensity returned to pre-order levels after controlling for those factors would weaken the conclusion that the protections remained effective. Evidence of sustained reductions across those matched measures would strengthen it.[1][3]

The immediate verifiable milestones are the consultation close on 16 November 2026, the final instrument, ministerial approval and the transition arrangements for any amendments. The deeper policy test is not whether a dramatic percentage can be repeated. It is whether the rules continue to reduce loss severity while issuers adapt their product design, settlement methods, incentives and margin systems around them.[1][4]

Report an error or submit an update