Liquidity sweeps: what the term actually describes

[Shadow Trader]

A liquidity sweep is a quick push through an obvious level, where stops and pending orders sit, followed by a reversal. Price takes out the previous high, grabs the resting orders, and snaps back. Social media treats it as a secret. It's closer to a description of a common shape.

The shape is real, since clustered stops do create fast moves when triggered. The trouble comes with using it as a trading rule. Every reversal after a spike can be labelled a sweep afterwards, and every continuation can be labelled a genuine breakout. Hindsight makes the label fit anything.

Testing requires a precise, forward-looking definition. For example: price exceeds the high of the previous 60 minutes by at least one pip, then closes back inside the range within three candles. Count every occurrence automatically and measure what happens over the next 10 and 30 minutes, after costs.

That kind of test often shows a modest tendency to revert in certain sessions and nothing in others.

A practical version of the test: take every 60-minute high and low over a year of one-minute data. Flag every instance where price pokes beyond by at least a pip and then closes back inside within five minutes. Record the maximum move in the original direction and the reversal distance over the next 15 minutes, net of a spread. The result is a table anyone can inspect, and it quickly separates the sessions where the shape pays from those where it doesn't.

Many find that sweeps around the London open show the strongest reversal tendency, while sweeps after major news show almost none. Whatever the outcome, the numbers beat the screenshots.

For anyone building the test, a useful extension is classifying each sweep by what happened next. Category one: clean reversal, price travels back through the range. Category two: partial reversal, then continuation in the direction of the sweep. Category three: no reaction, price simply keeps going. The proportions in each category, by session and pair, form a base rate that can be compared with any claim made online.

It also helps to record the context. Did the sweep happen at a level that many traders would watch, like a prior day high, or at an arbitrary swing point? Were spreads normal? Did a news event fall within the previous half hour? Tagging these details turns a vague concept into a dataset.

After a few hundred cases, the picture usually shows pockets where the shape has a modest edge and wide areas where it has none. Those pockets, tested on fresh data, are the only part worth trading.

Has the sweep concept ever been defined precisely enough to test, and what did the numbers say?