Last Look in Forex: Why a Liquidity Provider Rejects You

An order goes out at a price that was on the screen. It comes back refused. Nothing was wrong with the platform, and the price was not a mistake.

The party quoting it had a moment in which to decide whether to stand behind it. That moment is called last look, and it is why a visible price in the currency market is not always a price you can have.

What follows is what happens inside that window, which parts are governed by a published standard, and which parts each provider designs for itself as long as it says so.

What Last Look Actually Is

The FX Global Code defines last look as a practice in electronic trading where a market participant receiving a trade request has a final opportunity to accept or reject that request against its quoted price.

Two roles matter. The party streaming prices is the liquidity provider. The party sending a request to trade on one of those prices is the liquidity consumer.

The consequence sits in the word request. On a venue that operates a central limit order book, matched interest is binding, and the prices submitted there represent firm liquidity. Where last look applies, the streamed price is indicative and the provider retains discretion.

This is not a loophole bolted onto an otherwise firm market. Most electronic currency transactions are dealt this way, because providers stream quotes to many consumers at once and cannot know how many requests will arrive against the same quote, or how late one might turn up.

The trade-off is stated plainly by the Global Foreign Exchange Committee. Providers manage that exposure through the checks, and one consequence is the possibility of rejection, which can leave the consumer transacting later at a worse price.

The Two Checks Inside the Window, and Only One Has a Direction

Most explanations treat a rejection as a single event. Principle 17 does not. It describes last look as a risk control mechanism used to verify validity and price, and those are two distinct tests.

The validity check confirms that the transaction details in the request are appropriate from an operational perspective, and that there is sufficient available credit to enter into the transaction. It is a question about the request itself and about the credit line behind it.

The price check confirms whether the price at which the request was made remains consistent with the current price that would be available to the client. The provider compares the quoted price against its own estimate of the current price, and the permitted difference is expressed as a price tolerance, typically in pips, basis points or a percentage of the spread.

Only the second test can be influenced by which way the market moved. A validity failure is indifferent to price direction entirely.

Separating them matters because the two failures mean different things. A credit rejection is about limits and exposure. A price rejection is about latency and about how the provider has set its tolerance.

Symmetric and Asymmetric Price Checks

Under a symmetric price check, the tolerance is applied evenly around the current price. A request is rejected when the movement exceeds the tolerance regardless of whether that movement favours the provider or the client.

Under an asymmetric price check, the tolerance is applied unevenly. The Global Foreign Exchange Committee describes the tendency this produces: trades are accepted when the price has moved in favour of the provider, while requests are rejected when the price moves beyond the tolerance against the provider.

A related variant exists. Some providers apply a price improvement mechanism, accepting the request at a better price where the market has moved in their favour rather than rejecting it.

Here is where a claim in circulation needs correcting. Asymmetric price checking is frequently described as a breach of the Code. That is not what the Code or the Committee says.

The Committee’s position is that both methods exist in the market, and that the choice of price check logic could materially affect the predictability of the process and therefore the outcomes for the client. Its recommendation is a disclosure requirement: providers should disclose whether their price check is applied symmetrically or asymmetrically.

It adds that asymmetric checks make execution harder for consumers to monitor, which is why disclosing the background matters where they are used. The obligation attaches to transparency, not to the design itself.

What the Code Does Not Allow During the Window

Principle 17 gives the provider sole discretion over acceptance, based on the validity and price checks, and acknowledges that this leaves the client carrying market risk if the request is refused. Several restrictions follow from that.

Last look should not be used for the purposes of information gathering with no intention to accept the request to trade. A window used to see what a client wants, rather than to decide whether to fill it, is outside the principle.

Confidential Information arises at the point the participant receives the trade request, which is the start of the window rather than the moment of acceptance. Handling it is tied to the Code’s principles on information sharing.

Participants should not conduct trading activity that uses information from the client’s trade request during the window. The Code names two forms specifically: pricing activity on electronic trading platforms that incorporates information from the request, and hedging activity that incorporates it.

The stated reason is direct. Such activity risks signalling the client’s trading intent to other participants and could move prices against them, and if the request is then rejected the client is left worse off than before they asked.

Duration matters too. Because the consumer carries market risk for as long as the window lasts, providers should apply the checks without delay, and any delay beyond what those checks require is considered contrary to Principle 17.

Cover and Deal, and the Three Conditions That Define It

There is one carve-out from the restriction on trading during the window, and it is drawn narrowly. It is known as a cover and deal arrangement, and the Code sets out three characteristics that an arrangement must have in full for the guidance not to apply.

First, an explicit understanding that the participant will fill the client’s trade request without taking on market risk in connection with it, by first entering into offsetting transactions in the market.

Second, the volume traded during the last look window is passed on to the client in its entirety. Third, the understanding is appropriately documented and disclosed to the client.

All three are required, joined by and rather than or. An arrangement holding two of the three is not a cover and deal arrangement.

The logic is that the trading happening in the window is being done for the client rather than ahead of them, and the whole of it reaches them.

The Committee also recommends that providers disclose whether they source liquidity this way, and notes that a consumer will want to know when it applies, such as whether it is limited to particular currency pairs or hours.

Why No One Publishes a Standard Window Length

Articles on this subject routinely attach a number to the window. This page does not, and the reason is worth stating.

No official body publishes a standard, a maximum, or a typical last look window length. The figures that circulate are not traceable to the Code, to the Committee, or to a regulator, and a figure that cannot be sourced does not belong on a page like this one.

What the Code does require is disclosure by the provider. A participant should disclose, at a minimum, whether and how changes to price in either direction affect the decision, the expected or typical period of time for making that decision, and the purpose for using last look at all.

The Committee goes further in its recommendations, asking providers to disclose the maximum and minimum length of their window and to describe any circumstances in which those times may change.

That converts an unanswerable general question into an answerable specific one. There is no correct window length to look up. There is a disclosure that a given provider either publishes or does not, and the absence of one is itself informative.

Where a Retail Order Actually Meets Last Look

A retail client does not usually face a liquidity provider’s last look directly. The chain has a middle link, and confusing the two ends of it produces most of the misunderstanding on this topic.

A retail order goes to a broker. Depending on how that broker operates, it may then send its own request to one or more liquidity providers, and it is that request which meets the last look window.

This is why a requote and a last look rejection are not the same event. A requote is the broker’s response to the client’s order. A last look rejection happens one layer further out, on a request the client never saw. Our page on slippage, requotes and rejections separates the outcomes a client can actually observe; this page describes the mechanism behind one of them.

How much of this reaches a given client depends on the broker’s model, which our page on how broker execution models differ sets out. Where a broker prices internally rather than passing requests outward, no external last look is involved at all.

Venues using firm quotes, with no last look, also exist. The Committee notes this and frames it as something consumers can compare and choose between, much as exchange-traded and over-the-counter markets differ in whether a displayed price binds. The depth of market a platform shows carries the same caveat, since an indicative ladder is not a set of firm commitments.

Who This Page Is Not For

This page explains a mechanism. It does not identify a provider or broker that applies last look unfairly, and it offers no way to score one against another.

It contains no rejection rate, no window length and no acceptance statistic, because no official source publishes those figures and an unverified number would be worse than none.

Anyone looking for a verdict on whether last look is good or bad will not find one here either. It is a feature of an over-the-counter market with both a stated rationale and documented ways of being misused, and the Code’s response has been to require disclosure rather than to prohibit it.

The useful part is narrower. It tells a reader what to ask for, which is the provider’s own disclosure, and what to look for in it: whether the price check is symmetric or asymmetric, what the window length is, and whether liquidity is sourced through a cover and deal arrangement.

Anyone whose money sits with a broker rather than a provider will find where client money sits the more immediate question.

Frequently Asked Questions