Liquidity heatmap

How the picture is built

The input is a stream of order-book updates: for each price on each side, the total displayed size at that instant. The heatmap keeps a rolling history of those updates and draws each price as a horizontal band whose colour brightness follows the resting size. Time runs left to right, so the right edge is the current book and everything to its left is what the book looked like earlier. Trades are usually overlaid as dots, sized by executed volume, so the reader can see whether the size at a level was hit or simply disappeared.

• Colour scale: adaptive (auto sensitivity) or manual, with a contrast control; the same size should read equally bright on the bid and on the ask.
• Depth: only the levels the feed publishes are drawn. Depth received is a property of the feed, not a display setting.
• Resolution: one row per tick; vertical smoothing can blend neighbouring rows for readability at the cost of precision.

Reading a band

• A bright band that persists while price stays away is displayed size that nobody has tested yet.
• A band that fades as price approaches, with no trade dots at that price, is pulled liquidity: the size left without trading.
• A band that stays bright while trade dots accumulate on it is consistent with refilling, but several independent orders can produce the same picture.
• A band that ends exactly where a cluster of trade dots begins is size that was executed.
• A hole with no colour is either empty book or missing data; the feed's coverage decides which, and a coverage gap must not be read as zero.

Heatmap, footprint and DOM

The three displays read different datasets. The footprint aggregates executed volume per price per bar; the DOM shows the book at this instant; the heatmap shows the book through time. A vanished wall is only visible on the heatmap, a completed auction only on the footprint, and the queue in front of your own order only on the DOM. A liquidity heatmap is also distinct from a liquidation heatmap, which models where leveraged positions might be closed rather than displaying observed orders.

In Senzoukria

The Heatmap screen draws resting orders per price through time next to the trades that hit them. Its settings window groups controls under Rendering, Liquidity, Trades, Markers and DOM ladder: Intensity, Auto sensitivity or Manual scale, Contrast, Low floor (lots), Vertical smoothing, Volume dots with a volume filter, Best bid / ask lines, Cumulative depth and Highlight big orders. A Colour mode with Adds tint and Pulls tint separates liquidity being placed from liquidity pulled without trading, inferred from the aggregated book and labelled as such. On crypto, the Historical order book view loads recorded depth (Bybit, 200 levels per side, one-second samples) and leaves blank columns where there are no observations; on CME, live depth comes through a broker or Databento feed billed by the provider.

Common mistakes

• Trading a wall the moment it appears, before price has tested how the level behaves.
• Ignoring pulled liquidity. A vanished wall is an event worth recording, not a non-event.
• Comparing heatmaps from feeds with different depth or aggregation as if they showed the same book.
• Reading intensity across a colour-scale change: after retuning sensitivity, older bands are not comparable to newer ones.