Five challenges at 0.5 per cent each is not 0.5 per cent of risk. It is 0.5 per cent five times, on correlated instruments, against five different limits that reset at five different hours. The guard rails that keep a multi-account book inside all of them.
Five challenges running at once. Each one sized at 0.5 per cent per trade, which is conservative by any standard. On a Thursday in the illustrative example below, all five take the same gold short, gold runs 180 points against the idea, and every account takes a 0.5 per cent loss. The trader's actual drawdown that morning is 0.5 per cent of five accounts, taken at the same minute, on one decision that turned out to be wrong.
The answer to the question underneath that scene: sizing correctly on each account is necessary and it is not sufficient. Multi-account risk management for prop challenges is a different discipline from single-account sizing, and it needs guard rails that operate above the individual trade — a daily halt that is enforced per account, sizing computed from each account's own equity, a cap on correlated exposure across the whole book, and an awareness that five firms have five different reset hours. This post is about those guard rails specifically.
INFO
Quick answer. Four guard rails keep a multi-account prop book alive: a per-account daily halt that stops that account trading when its own loss limit is approached, sizing computed from each account's own equity and the actual fill price, a correlated-exposure cap counted across all accounts rather than per trade, and a reset-hour map so you know when each firm's day actually begins.
Ask a trader running several challenges what their risk per trade is and you get a per-account answer: half a per cent, one per cent. Ask what their risk per decision is and the room goes quiet, because that number is the per-account figure multiplied by the number of accounts taking the trade, and nobody has written it on anything.
Worked as an illustrative example, with five accounts at 0.5 per cent:
Nothing in that arithmetic is exotic and nothing is wrong at the account level. Every account is inside its per-trade rule the whole way down. What has happened is that the book has an exposure profile no individual account can see, and no per-account rule can control.
The consequence is a specific and common failure: a trader passes four challenges, fails all four in the same week, and concludes their strategy stopped working. The strategy did not change. The correlation did.
Every prop firm has a daily loss limit. It is the rule that fails more challenges than any other, and the reason is that it is a hard floor that does not care what you intended.
A daily halt is the mechanism that stops an account trading when it approaches its own limit, before the limit is touched. Three properties make it work.
It is per account, not per book. Account C being down 3.2 per cent must halt account C. It must not halt accounts A, B, D and E, which are down 0.4 per cent and are having a normal day. A halt that applies to everything at once turns one bad account into five stopped accounts.
It fires on a threshold below the limit, not at it. If the firm's daily limit is 5 per cent, halting at 4.9 per cent is theatre — a single open position moving against you crosses the last tenth before anything can act. A halt at 3.5 or 4 per cent leaves room for open positions to be closed in an orderly way. The gap between the halt and the limit is the room your open risk needs, so size it against your largest normal position, not your average one.
It counts the way the firm counts. This is the part that catches people. Some firms compute the daily loss from the balance at the reset hour, some from the highest equity reached that day, some include open floating loss and some do not. A halt that measures a different number from the firm's is a halt that fires at the wrong time in both directions.
WARNING
A daily halt that only blocks new trades is half a guard rail. If four positions are open when it fires, the account can still breach its limit while the halt is active. Decide in advance, in writing, whether the halt also flattens open positions — and know that flattening has its own cost on a position that would have recovered.
The rule is short. Every receiving account computes its own position size from its own current equity and the actual distance to the stop, measured against the price it actually filled at. It does not copy the Provider's volume and it does not scale from a fixed ratio.
The arithmetic of that, and what happens when lot steps and minimum volumes get in the way, is a whole subject in itself and is covered separately. What belongs here is the multi-account consequence: sizing per account is what makes the accounts comparable. If all five accounts take the same percentage risk on every trade, then five equity curves that diverge are telling you something real — one account is on a different broker with different spreads, or one account skipped trades it could not size. If the accounts are sized by different methods, divergence tells you nothing at all, because you cannot separate the strategy from the configuration.
One practical detail that is easy to get wrong: measure the risk percentage on the fill, not the signal price. On a challenge account, a systematic few points of slippage on entry becomes a systematic overshoot of your intended risk, on every trade, on every account, for the whole challenge.
This is the guard rail that most multi-account setups simply do not have, and it is the one that produces the "failed four at once" week.
The mechanics are straightforward once stated. Your exposure limit should be expressed as a total across the book, on correlated groups, not per position and not per account. Something like: at most two open positions in the dollar-correlated group, across all accounts, at any time.
Building a usable correlation map does not require statistics. A rough grouping is enough, and rough is what you will actually maintain.
Two rules that use that map:
The grouping matters most in exactly the situation where it is hardest to remember it: a strong trending session, when several setups fire at once, all pointing the same way, and every one of them looks independently justified.
Five firms, five definitions of "today". This sounds like a footnote and it is the source of a specific, avoidable failure.
If firm A resets its daily loss counter at 00:00 server time and firm B resets at 17:00 New York, then a loss taken at 16:00 New York sits in different days on the two accounts. Stop trading because account A has hit its daily halt, and account B may have just started a fresh day with a clean counter — or the reverse, where a trade you consider the first of the new day is, on one account, the fifth loss of a day that has not ended yet.
The fix is administrative rather than clever. Write the reset hour of every account you hold in one place, converted into your own local time, and re-check it at the daylight-saving changeovers when server times move relative to each other. Firms' reset definitions as of September 2026 vary widely, including between products at the same firm, so read the current terms per account rather than assuming a house standard.
Being plain about the limits, because the marketing version of this subject is not.
They cannot prevent a gap. A position held over a weekend or into a release can open beyond its stop, and no halt, cap or sizing rule acts between the close and the open.
They cannot make correlated bets uncorrelated. A cap reduces how many you hold. It does not change the fact that the ones you hold move together.
They cannot promise you pass anything. A guard rail is a constraint on losses you control. Challenges are also failed by ordinary losing streaks inside the rules, and no configuration removes that.
They cannot substitute for reading the terms. Every number in a guard rail comes from a firm's rulebook, and the rulebook is the firm's to change.
Fewer than the number that makes your risk per decision uncomfortable. If you take one trade and it reaches every account, your real exposure is the per-account risk multiplied by the account count — five accounts at 0.5 per cent is 2.5 per cent of your aggregate capital on a single idea. Decide the aggregate figure you are prepared to lose on one decision first, then divide to get the account count.
A daily halt stops an account opening new trades once its own loss for the firm's trading day reaches a threshold you set below the firm's hard limit. It must be per account because accounts have different balances, different limits and different reset hours, and because one account having a bad day is not a reason to stop four accounts that are trading normally.
Count open positions by correlated group across the whole book rather than per account, and cap the group. Three accounts each holding one index trade is three index positions in one group. A rough grouping — dollar pairs, metals, indices, yen crosses, energy — is enough, and it is the one you will keep updated, unlike a correlation matrix.
No, and the differences matter. Some measure from the balance at the day's reset, some from the highest equity reached during the day, and firms differ on whether open floating losses count towards the limit. Reset hours differ too. As of September 2026 there is no house standard — check the specific terms for each account you hold, including between products at the same firm.
No, by default. Halting the account that is in trouble is risk management; halting the others is an unrequested change of strategy on accounts that are inside their rules. The exception is a deliberate book-level circuit breaker, set by you, for the case where the whole book is down — and that should be a separate, explicitly configured rule rather than a side effect.
That is a decision you make in advance, and both answers are defensible. Blocking new trades while leaving positions open preserves trades that may recover but allows the limit to be breached by the open risk. Flattening everything caps the day at a known number and realises losses that might not have been realised. What is not defensible is not knowing which one your setup does.
The per-trade arithmetic that sits underneath guard rail two — lot steps, minimum volumes and what to do when a size rounds to nothing — is worked through in /blog/prop-firm-copier-sizing-across-accounts. And because guard rail one is only as good as your understanding of the number it is watching, /blog/what-is-daily-drawdown covers the three ways firms compute a daily loss and why the differences change when your halt should fire.
