If-Then Orders in FX and Crypto: How OCO and Brackets Really Behave

A simple idea with a serious job

Imagine you are long a position. You have a profit target above the market and a protective stop below it. Two orders, one hope, one fear. If the target fills, you are out. But that stop is still sitting there, and it does not know you have left. If price later falls to that level, it triggers and sells something you no longer own, quietly opening a short position you never asked for. This is the exact scenario OCO exists to prevent. OCO stands for one-cancels-other: two orders linked so that when one executes, the other is automatically cancelled.

That linkage, and not the order tickets themselves, is the entire feature. Two unlinked orders are just two orders. The CFTC's glossary describes OCO as a pair of orders, typically limit orders, where a fill on one automatically cancels the other. Regulated brokers describe the same mechanic, usually pairing a take-profit limit with a protective stop on a position you already hold.

OCO, OTO, and the bracket family

Three structures get bundled together in conversation, so it helps to separate them. OCO links two orders for a position that already exists, with no entry component. OTO means one-triggers-other: a primary order that, once filled, submits one or more secondary orders that did not exist before. A bracket order combines them, pairing an entry order with an OCO exit pair, a profit target above and a stop below for a long trade. That is why brackets are sometimes described as OTO plus OCO.

The practical difference matters. A bracket defines your whole trade in one ticket before any capital is committed. A standalone OCO adds exits after a manual entry, which means a separate step, and any separate step is a step that can fail to happen. Trading Technologies' documentation describes its bracket as launching an OCO child order for each partial or full fill of the entry, with profit and stop offsets set from the fill price.

Who is actually holding your order?

Here is the part most tutorials skip. In futures, CME Globex does not natively support OCO, so the linkage is enforced by your platform or broker, not the exchange. In spot FX and crypto, the mechanics are venue-specific, and many venues manage conditional orders in their own systems rather than as a single exchange-native instruction. The CFTC definition covers US regulated futures and options, and the documented broker examples cover stocks and ETFs. None of those sources specifically documents spot FX behaviour, so treat claims of identical behaviour across venues with caution. Check your own platform's documentation.

That distinction has real consequences. Does your linked pair survive if you close the app or lose connection? Is it a day order or good-till-cancelled? Does it work in sessions where the venue accepts only limit orders? Which price triggers the stop leg? One platform's helpful automation is another's surprise.

Partial fills are where predictability ends

Partial fills are the least predictable corner of linked orders, and behaviour differs by platform. Suppose you plan 300 shares and only 100 fill. Many platforms size the exits to the 100 you actually hold; others place exits for the full planned size or wait until the entry completes. Take the Trading Technologies example: a 10-lot bracket partially fills for 7 lots, so a 7-lot OCO is submitted. If the profit leg then fills 3 lots, the stop is reduced by 3. Sensible, but not universal. Some systems cancel a stop the moment any part of the target fills, which can leave part of your position unprotected. Check your position after every partial fill.

There is also the rare double-fill: in very fast markets, both legs can execute before the cancellation reaches the other side, leaving an unintended opposite position. Rare, but real rather than theoretical. And cancelling a linked group often needs the venue's dedicated complex-order cancel, because an ordinary cancel can dismantle half a bracket without any error message. A stop removed while a target stays live is the kind of silence you do not want.

What OCO guarantees, and what it does not

OCO guarantees that one order is cancelled when the other fills. It guarantees nothing about price. A stop remains a stop: FINRA is explicit that a stop price is not a guaranteed execution price, and stop-limit orders may never execute at all. If markets gap through your level overnight or after a halt, your fill can land well beyond your intended exit. A bracket defines your planned loss; it does not enforce it. Phillip Nova notes that for certain CME Group products, stop-market orders may be converted to stop-limit orders on trigger as part of exchange price protection. Useful to know before you assume a market exit.

In fast moves, prices can also touch your stop before reversing, and because the target was cancelled, you will not participate in what follows. The structure did its job; the market simply had other plans.

The quiet spread of features

Availability is uneven. Some brokers offer native brackets; others require an expert advisor or advanced ticket; some crypto venues expose brackets in the interface while others only via API layers. Time in force must usually match across an OCO pair, so you cannot pair a day order with a good-till-cancelled one. For multi-session protection, good-till-cancelled is often the tool, though overnight margin requirements are typically higher and weekend gaps can be unkind. And OCO does not size your position. Sizing decides what a bad exit costs, which is upstream of every order type here.

None of this makes linked orders bad. It makes them worth understanding before you lean on them. A bracket is a plan written down in advance, which is usually better than a plan improvised at the loudest moment. Just confirm, on your own venue, who is holding the order and what happens when the connection blinks. The market will not send a reminder.