Order Execution Latency: Where Time Is Lost Between the Click and the Trade

Execution latency is the time between the moment you press "buy" and the moment your order joins the queue in the exchange order book or finds a matching counter-order there. It is made up of legs for which different parties are responsible: your own connection, the broker's gateway with its mandatory checks, the exchange gateway and the matching engine, and then the return path of the execution report. What matters to a trader is not the time as such but its consequence: the exchange assigns queue priority by price and by time of arrival, so an order that arrives late stands behind those that got there earlier and is filled at a different price — or is not filled at all.

What the order's path is made of

The first leg is on the client side: the terminal builds a message and sends it to the broker. Here the time depends on the connection, on how heavily loaded your device is, and on whether you trade through a mobile app, a desktop terminal or a trading API.

The second leg belongs to the broker. Before passing the order on to the exchange, the broker is obliged to run pre-trade controls: to check whether there is enough collateral, whether the order breaches any limits, and whether the security is permitted for your risk level. This check is carried out for every order and is part of the normal process, not a malfunction.

The third leg is the exchange. The order is accepted by the gateway, reaches the matching engine and either meets a suitable opposite order or takes its place in the book. This is where latency is least within a retail trader's control and most even for everybody.

The fourth leg is the way back. The execution report travels to the broker and on to the terminal. It is this leg that causes the most common misunderstanding: the trade has already happened, yet you still see the order as active and have time to send a second, duplicate order.

A separate source of discrepancy is the latency not of the order but of the data. If quotes arrive late, you make your decision on a price that is no longer in the book. Formally the order left instantly; in practice it was late before it was even sent. That is why it helps to distinguish "slow to execute" from "slow to see" — these are different problems with different solutions.

Why latency turns into money

For a market order, latency means slippage: you get whatever price is left by the time the order arrives, and in a fast market that price differs from the price you saw. The thinner the book, the more your own order moves the price — more on this under the term price impact and in the article on how an order affects the price.

For a limit order, latency means something else: you do not get a worse price, but you lose your place in the queue. While the order was in transit, the best levels were taken out, and your price ended up away from the market. This is where a partial fill comes from — part of the volume was taken, the remainder was left hanging, and from then on the fate of that remainder is decided by the order's time in force.

The overall effect is best counted not in seconds but in money: the difference between the decision price and the price of the actual trade is precisely the execution cost, in which latency sits alongside the spread and the commission.

When latency is noticeably longer than usual

There are periods when the wait lengthens for good reason. The market open and close, auctions, the release of an issuer's results, corporate actions, sudden news — all of these increase the flow of orders and widen the spread at the same time. It is worth checking the timetable of such moments in advance: the events calendar, company reports and the news feed.

A halt under exchange rules is a case apart. If the price moves beyond the set limits, the exchange switches the instrument to a discrete auction or suspends trading. From the terminal this looks like a stuck order, but there is no technical latency here — there is a rule under which no trades are concluded at that moment.

The liquidity of the instrument also changes the picture: in securities with a deep book, such as those gathered in the stocks section, a counter-order is almost always there, whereas in rarely traded issues an order may wait a long time for a counterparty — and that is no longer transmission latency but the absence of a second side to the trade.

What a trader should do

Step 1 — separate the diagnoses. Compare the time the order was sent, as recorded in the terminal log, with the time of the trade in the broker's report. If they are close and the price is still worse than expected, the cause is not latency but the spread and the depth of the book.

Step 2 — do not send a duplicate until the report has arrived. Clicking again on an unconfirmed order is the most expensive mistake in this whole subject.

Step 3 — choose the order type to fit the task. If you need certainty that the trade happens, use a market order, and then you pay for latency in price. If you need certainty of price, use a limit order, and then you pay with the risk of non-execution.

Step 4 — do not fight for speed where the time horizon is what decides. In intraday strategies fractions of a second matter; in positions held for weeks, execution latency usually matters less than the choice of entry level and the size of the position.

What latency does not explain

It is often blamed for any inconvenient result. But if an order was not filled at a price that "traded on the chart", the cause usually lies elsewhere: the chart is built from trades, while an order is filled against the book, and the available volume may simply never have reached your level. If a position was opened at a worse price in a thin instrument, it was your own order working against a sparse book. And if the order had disappeared by the next day, its time in force had expired. It is useful to read the terms in sequence and check the wording against the glossary: a cause that has been named can be fixed, a cause written off as "lag" cannot.