Liquidity Sweeps and Book Imbalance: Read the Measurements

What EdgeX calls a sweep pattern

EdgeX looks for same-direction executions within a completed window, spanning the configured number of ticks across at least three distinct prices, progressing monotonically in the aggressor direction. The observation may contain multiple participants. The detector does not claim a shared parent order or execution-chain identity.

That definition is narrower than an informal chart label such as liquidity grab. It gives you something measurable to review: direction, price span, number of traded prices and time window. Whether a particular pattern has predictive value is a separate research question that cannot be answered by the marker itself.

A volume burst does not identify stop orders

A burst means directional executed volume exceeds a configured floor in the observation window. It can occur during many kinds of participation. The trade prints supplied to this feature do not identify which executions originated from stop orders, so labeling every burst a stop run would add information that is not present.

Keep the neutral measurement in your records: how much classified volume, which direction, how long and how far price traveled. If you also write a market narrative, separate it from those facts so that another reviewer can understand what was measured and what was inferred.

Calculate normalized book imbalance

EdgeX uses 100 × (bid quantity − ask quantity) / (bid quantity + ask quantity) over the selected available levels. Positive values mean more displayed bid quantity; negative values mean more displayed ask quantity. A marker appears when the absolute percentage crosses the configured threshold.

The denominator makes this different from a simple bid-to-ask ratio. The selected depth range also matters: a near-market imbalance and an imbalance over many levels need not agree. If fewer levels are available than requested, the measurement describes the available sample.

Understand what a liquidity-change event includes

The detector compares only prices present in both consecutive snapshots within the selected depth range. It omits newly appearing or disappearing boundary levels, helping avoid mistaking a shifted viewing range for a quantity change at a fixed price. Both the net-quantity and percentage thresholds must pass.

CME’s order-book methodology describes updates arising from new orders, executions and cancellations. EdgeX’s net change between samples can include these effects together. It is therefore not a pure counter of added or canceled orders, and should not be described as one.

Reference: CME Group: order-book updates and liquidity measures

Use the Activity view to keep interpretations honest

Open Plugins, enable only the detectors relevant to your question, and inspect observations in Activity. EdgeX retains up to 100 observations from the last 60 seconds relative to chart time. That recent review is useful for inspecting evidence; it should not be mistaken for a permanent, complete research archive.

Write down the thresholds and environment before the session. Compare the marker with price response and current liquidity, and include data gaps in your record. Stale depth clears current results, and short-lived activity can be missed by sampling. The advantage of putting these measurements beside the chart is a clearer review process, not a promise that several markers together produce a reliable entry.

Common questions

Continue with the EdgeX product guides

These guides document the controls and data behavior used in this article.

• Read sweeps, bursts and liquidity changes
• Calibrate Order Flow for your instrument
• Order-flow plugins: signals, thresholds and evidence