What a sweep looks like on a heatmap

A sweep on a heatmap is the picture left when aggressive orders trade through several price levels at once. A heatmap draws resting limit orders as colour bands across time. A sweep ends a run of those bands in the same instant. This note walks through one invented ES sweep in Liquivue and lists what the heatmap shows and what it cannot show. The note on how trade bubbles sit on top of a liquidity heatmap covers the bubbles in detail.

What a sweep is

A sweep is a burst of market orders, or marketable limit orders, that takes all the resting size at one price and keeps going to the next. One large order can do it. So can many small orders that arrive within a few hundred milliseconds. The exchange fills each price in turn until the buying or selling stops.

The book before the sweep

All numbers below are invented. The market is the E-mini S&P 500 future, ES. One tick is 0.25 points. At 10:15:20 New York time the best bid is 5,039.75 and the offers look like this.

On the heatmap these are five bands above the price line. The band at 5,041.00 is the brightest, because it holds the most contracts. The four bands below it are dimmer.

The sweep, contract by contract

At 10:15:21.400, buy orders for 260 contracts arrive within 300 milliseconds. The exchange fills them against the offers from the lowest price up.

Four price levels empty out, and 30 contracts trade at the fifth. The last price moves 4 ticks, from 5,040.00 to 5,041.00, in 0.3 seconds. The average fill for the buyers is 5,040.55, about 2.2 ticks above the first offer.

What the heatmap draws

The heatmap shows three things at 10:15:21.

1. The bands at 5,040.00, 5,040.25, 5,040.50 and 5,040.75 all stop in the same column.
2. The band at 5,041.00 continues, but its colour drops a step, because its size fell from 210 to 180.
3. Trade bubbles print in a diagonal line up through the five prices. Their sizes are 30, 45, 60, 95 and 30 contracts.

Cumulative delta, or CVD, adds buy volume and subtracts sell volume. Here CVD jumps by 260 contracts in under a second, because every contract in the sweep bought at the offer.

A sweep compared with pulled liquidity

Suppose the traders at 5,040.75 had cancelled their 95 contracts instead. That band would still end, but no bubble would print at 5,040.75. CVD would not move.

A band that ends where bubbles print at its price traded away. A band that ends with no bubbles came from cancelled orders. The note on how long resting orders last at each distance from price shows how often bands near price end through cancels.

The same sweep in Liquivue

In Liquivue, the heatmap layer draws the five offer bands, and trade bubbles print at each level as the buy orders fill. Liquivue lists sweep markers, so the platform can flag the burst on the chart. The CVD panel steps up by 260 contracts at 10:15:21. The volume and delta profiles add 260 contracts of buying across the five rows. The VPOC moves only if one of those rows now holds more volume than the old VPOC row.

Liquivue runs in a desktop browser with WebGPU and connects to futures through Rithmic or the NinjaTrader bridge. It costs €20 a month after a 7-day trial, and your data provider bills market data separately.

What the heatmap cannot tell you

The heatmap cannot tell you whether one trader sent 260 contracts or twenty traders sent 13 each. Market by price data shows the total at each price, not each order.

It also cannot show hidden size. If 120 contracts traded at 5,040.75 while the band showed 95, the extra 25 came from an order the book did not display. Liquivue lists iceberg detection for that case, but the feature is experimental and needs live CME order-level data.

Sweeps on other platforms

Bookmap draws a historical order book heatmap with volume bubbles, and its add-ons cover stops, icebergs and other order flow events. Live futures analysis starts with its Global tier at US$49 a month. NinjaTrader Order Flow+ has a market depth map and cumulative delta inside the NinjaTrader desktop platform. The rankings page compares heatmap coverage across platforms.

A check you can run

Find three moments in your last session where price moved 4 ticks or more within one second. For each one, count the bands that ended in that second and the bubbles that printed at those prices. Write down how many bands traded and how many were pulled.