Can an intraday dip breach an end-of-day drawdown? On most desks, yes

An end-of-day drawdown is a floor sitting a published distance below your balance, and it moves upward only after a higher balance settles at the daily close. That is a movement rule. It fixes where the line sits today and says nothing about the moment the line is read.

The reading is a second rule with its own sentence: at what point does the firm compare your account to that floor, continuously, against equity that includes open positions, or once, against the settled close? A plan can publish the first sentence clearly and never write the second, which is why the two facts have to be collected separately rather than inferred from one label.

We sourced the reading rule for all eight covered firms across both phases on 2026-08-24, one official quote per cell. Six of the eight publish live enforcement somewhere in their own terms; three firms account for the five cells that publish no timing sentence at all. Not one covered firm publishes close-only checking, so nobody has earned that state in our data.

• Movement: what makes the floor rise. A settled higher close, and how much of the gain counts.
• Enforcement: what triggers the comparison, a live equity read including open trades, or the settled closing balance.
• Silence: a plan that answers the first question and skips the second has answered one of the two facts you need before sizing.

The wrong move is to defend the closing balance. A trader who has read that the floor only ratchets on a settled close reads that as room to let a position breathe, on the reasoning that the round trip ends flat or better by 4:00. Under live enforcement the round trip is never completed: the platform compares equity to the floor while the trade is open, and the first tick through the line liquidates the account.

The worked example below lands $300 through the floor on a day that closes $2,400 clear of it. That $300 costs the whole account, and the $2,400 arrives hours after the comparison has already fired. Recovery is irrelevant to a rule that never waits for it.

Which number gets compared matters as much as when. Where firms state it, the comparison reads equity (realised balance plus the profit and loss on anything still open) so an unrealised loss counts at full size while the trade is still yours to manage. A mental stop you intend to honour is invisible to that read.

Read as a claim about generosity, the model inverts. End-of-day movement really is the more tolerant half: an intraday spike to a new high tightens nothing, so today’s room is set by yesterday’s settled close and can only be enlarged by another settled close. All of that tolerance lives on the upside of the ratchet.

The downside carries no matching concession. The same account that refuses to raise the floor for an intraday high will still measure an intraday low against it, and where firms publish both sentences that is exactly the pair you find. The asymmetry is the design: peaks are ignored, troughs are counted.

In the worked example the floor could only have moved on the $51,800 settled close, up to $49,800 for the next session. The $49,100 equity low was measured against the standing floor the moment it printed. One number waited for the close; the other did not.

The wrong move with a silent page is to fill the gap with the industry’s habit. Five of the fifteen end-of-day cells we sourced carry no enforcement sentence anywhere: the firm publishes what the floor is and what moves it, then never states what triggers the check. Silence tells you about the page and carries no information about the rule.

Price that silence as live. If the check turns out to run on the close, assuming otherwise costs a little room and nothing else; if you assume the close and the check is live, the mistake costs the account and the fee that bought it. Asking support in writing is worth doing, and worth filing separately. A reply in a ticket is not a published term, and only the published term binds the firm.

“Not published” is a complete answer to give yourself. Record it in your own plan notes as its own state, distinct from “checked at the close”, because the two read almost identically on a screen and behave nothing alike on a bad tick.