On 15 January 2015 the Swiss National Bank abandoned the floor it had been defending under the euro against the franc. The pair moved further in minutes than it normally moves in a year, and it moved through a vacuum: there were no bids at the levels where retail stop losses were sitting. Accounts did not close at their stop-out level. They closed far below it, and a large number of them closed below zero.
That is the scenario negative balance protection exists to answer. Not an ordinary losing streak, which the account simply absorbs, but the discontinuous move that leaves the arithmetic broken.
What the rule actually promises
The promise is narrow and precise. A retail client's liability arising from leveraged trading is capped at the funds in the relevant trading account. If the account goes negative, the firm writes off the shortfall rather than issuing a demand for it.
Everything the promise does not cover matters just as much. It does not protect the deposit, which can be lost in full. It does not stop the position from being closed. It does not apply to the client's other accounts, and it says nothing about whether the firm can pay, which is a separate question answered by client money segregation and by capital rules. Leveraged trading remains high risk with the protection in place, and anyone told otherwise has been sold something.
Per account, and why that boundary is deliberate
The European approach applies the protection on a per account basis. All positions and cash inside one trading account are netted, then the protection is assessed on the result. It is not applied per position, which would let a client structure a book where one leg is protected and the other is not.
The consequence people miss is the reverse case. A client with three accounts at the same firm cannot have a profit in account two used to plug a hole in account one; the accounts stand separately. Firms that let clients open many sub-accounts should expect that boundary to be tested after any violent session, and the account model in the platform has to reflect it before the event rather than after.
The rest of the package it arrived with
Negative balance protection did not arrive alone. The European product intervention measures that made it mandatory for retail CFD clients came bundled with leverage caps by asset class, a standardised risk warning, a ban on trading bonuses, and a margin close-out rule requiring firms to close positions once account equity falls to half of the required margin. The United Kingdom and Australia adopted equivalent packages for retail clients.
Those pieces work together. The leverage cap reduces the size a retail client can hold against a given deposit. The 50 percent close-out rule forces the exit earlier than a firm might otherwise choose. Both of them cut the frequency and size of negative balances, which is what makes the write-off promise economically survivable. A regulator handing out the protection without the caps would be handing brokers an unpriceable liability.
Automatic close-out is not a guarantee of the close-out price. In a gap there may be no counterparty at the trigger level, so the position closes wherever liquidity reappears. The protection covers the deficit; it does not prevent the loss.
Who absorbs the deficit
The firm does, and how much that hurts depends on how the flow was handled. A firm running client positions internally keeps the loss it would otherwise have booked as profit, so the negative balance is a direct write-off against its own capital. A firm passing the flow to a liquidity provider owes that provider the full amount regardless of what it can collect from the client, so the write-off is a real cash outflow into a market that has just gone against everyone. This is one of the sharper edges of the A-book and B-book decision, and it only shows up on the worst day of the decade.
Firms price the exposure in ways clients can see if they look. Reduced leverage on instruments with jump risk, wider margin requirements before scheduled events, higher stop-out thresholds, and restrictions on holding certain instruments over a weekend are all negative balance protection being paid for in advance. So is a firm's decision to avoid pegged currencies entirely, which several did after 2015.
Where the protection does not reach
Three gaps come up constantly in client complaints. The first is client categorisation: elective professional clients generally give up the retail package, higher leverage included, and many do not read the paragraph that says so. The second is jurisdiction. A firm licensed offshore has no obligation to offer the protection unless its own regulator requires it, and a contractual promise on a website is worth exactly what the entity behind it is worth, which is why checking which entity holds the account is worth ten minutes.
The third is product scope. The measures were written for CFDs sold to retail clients. Spot crypto on an exchange, futures accounts under other regimes, and prop firm evaluation accounts, which are usually simulated and governed by a contract rather than by client money rules, all sit outside them. A prop trader breaching a drawdown limit is not experiencing negative balance protection; they are experiencing the terms of their evaluation agreement, which is a different thing with a different remedy.
What a firm has to build
Operationally the requirement is a risk engine that values every open position against live prices continuously, applies the close-out threshold at the account level, and cannot be delayed by a queue during exactly the minute when everything is moving. It also needs a defensible audit trail: the tick used, the timestamp, the equity calculation, the order sent. When a client disputes a close-out, that record is the whole case.
Then there is the write-off workflow itself. Someone must reset the balance to zero, log the reason, and report the loss internally. Firms that handle this manually in a spreadsheet after a crisis session discover how many accounts were affected only slowly, and slow is expensive when the finance team is trying to size a hole.
"Every broker says they have negative balance protection. The question I would ask is what happens at the stop-out level on a Sunday open, because that is the moment the promise costs money."
— Roman Onta, Executive Director, SINGUARD
Key Takeaways
- The protection caps a retail client's loss at the funds in that trading account and does nothing to protect the deposit itself.
- It is assessed per account, so balances in a client's other accounts are neither used to cover a deficit nor exposed to one.
- Leverage caps and the margin close-out rule are what make the write-off affordable, which is why they arrived in the same package.
- Professional categorisation, offshore entities and prop firm evaluation accounts all sit outside the retail protection, whatever the marketing page says.
Frequently Asked Questions
What does negative balance protection actually guarantee?
It limits a retail client's loss on leveraged trading to the funds in that trading account. If a gap pushes the account below zero, the firm writes the deficit off instead of pursuing the client for it. It does not protect the deposited funds themselves, which can still be lost entirely, and it is a rule of specific regulators rather than a universal feature.
Is negative balance protection applied per position or per account?
Under the European approach it is applied on a per account basis, so all positions and cash in one trading account are netted before the protection is assessed. A client holding several accounts at the same firm does not get profits in one account used to cover a deficit in another, which is why the account boundary matters.
Do professional clients get negative balance protection?
Not automatically. The retail protections that came out of the European product intervention measures attach to retail categorisation, and clients who elect professional status generally give them up along with the leverage caps. Some firms extend the protection to professional clients as a commercial choice, but it is a contractual promise at that point, not a regulatory one.