Footprint Charts Explained

What is a footprint chart?

A footprint chart takes each candle and splits it into price rows, showing how much aggressive buying and how much aggressive selling executed at every row. Instead of one bar with one volume number, you get a column of cells: the price axis is preserved inside the bar, and every cell carries the two sides separately. It is the most direct visual form of order flow on a normal price chart.

Everything a footprint shows already existed in the exchange's trade stream. Each print carries price, size, timestamp and a maker flag from which the aggressor side is derived. Building a footprint means bucketing that stream twice — by time into bars, and by price into rows — and summing signed size in each cell. Nothing is modelled, smoothed or predicted. It is an exact reorganisation of published data.

Why bother? Because a candle answers "where did price go" and a footprint answers "at which prices did the volume actually change hands, and who was in a hurry there". A bar that closed on its high tells you nothing about whether the buying happened at the low of the bar or was chased at the top. Those are different events with the same candle. The footprint separates them, and that separation is what makes level-by-level analysis possible instead of pattern-matching on shapes.

It is also the surface where absorption becomes visible as an arithmetic fact rather than an impression: heavy aggressive volume concentrated at one extreme of the bar, and a close that came back the other way.

How do you read a footprint cell?

Each cell is one price row, and the standard bid/ask layout prints two numbers per row: the volume that traded on the bid — sellers crossing down into resting bids — and the volume that traded on the ask, buyers lifting resting offers. Read left to right, the pair tells you who was the aggressor at that specific price during that bar.

From those two numbers everything else in the display is derived:

• Row delta is ask volume minus bid volume: the net initiative at that price.
• Row total is their sum: how much business was done there at all.
• Point of control (POC) is the row with the highest total: the price this bar actually agreed on.
• Bar delta is the sum of all row deltas, which is the single number a CVD series accumulates.

The reading discipline is to check the three in order — where is the POC relative to the bar's range, where is the delta concentrated, and where did price close. A bar whose POC sits at the top and whose buying is concentrated at the top but which closes at the bottom is buyers getting filled at the worst price in the bar; the same bar with the POC at the bottom and the close at the top is the opposite. The candle body is identical in both cases.

One practical caution: cell counts get sparse fast. On a low-timeframe bar in a quiet market a row may contain one or two fills, and single-fill rows produce dramatic-looking ratios that mean nothing. Footprint is a high-density instrument; when the density is not there, the display keeps drawing anyway.

The four common footprint layouts

Footprint is a family of displays over the same underlying cell data, and each layout trades information for legibility. They are not different indicators — they are different renderings of the identical bid/ask matrix.

The workflow most desks land on is to keep a low-information layout up for context and switch to bid/ask only at the bar and the level under examination. The Confluence Engine follows the same logic: a compact footprint layer for scanning and a separate detailed footprint layer for inspection, with the same numbers underneath both so the two can never tell different stories.

Note that the profile layout is the bar-level cousin of a session volume profile. Same arithmetic, different aggregation window — and once you see that, POC migration from bar to bar reads naturally as a running answer to "where is this auction agreeing now".

What is a stacked imbalance?

An imbalance is a cell where one side's aggressive volume dominates the other by more than a chosen ratio. A stacked imbalance is several of those in a row at consecutive prices, in the same direction — the sign that one side worked through a series of price levels rather than dominating a single one by chance.

The important mechanical detail is that the comparison is normally diagonal, not horizontal. Buyers lifting the offer at price P are compared against sellers hitting the bid one row below, because those two orders sat opposite each other in the book. Comparing the two numbers on the same row compares trades that were never competing for the same liquidity. Implementations that do it horizontally will disagree with implementations that do it diagonally on the same data, and both will look plausible.

Two guards separate a real stacked imbalance from an artefact:

1. A ratio threshold. A common default is three to one. Lower it and every bar lights up.
2. A minimum volume floor on the aggressor leg. If the opposing cell is empty, the ratio is mathematically infinite and a single dust-sized fill would qualify. The Confluence Engine applies the floor unconditionally — before the ratio test, not only in the empty-cell branch — because a dust-on-dust pair can clear a three-to-one ratio on its own.

A stacked imbalance is a description of what was consumed, not a prediction. Its most useful property is that it leaves a marked price band you can watch on the retest: if that band is defended again, it was liquidity that mattered; if price passes back through it without reaction, it was liquidity that has already been spent. That failed-retest case is tracked as its own layer, because the failure is often more informative than the original stack.

Why do footprint charts look different on different platforms?

Because a footprint is not a single well-defined object: it is the output of at least four choices, and platforms make them differently. Same exchange, same asset, same timeframe — different picture, with no one being wrong.

The four choices:

• Row binning. Some tools use a fixed tick size per row; others divide each bar's range into a fixed number of bins, so row height changes with bar range. The Confluence Engine bins each bar's range into a fixed count, which keeps rows legible at any volatility but means the price height of a row is not constant between bars.
• Aggressor classification. Reading a published maker flag is a measurement. Estimating from the quote is a model. Bars built the second way will disagree in fast markets.
• Trade aggregation. Many exchanges publish aggregated prints that merge one taker order's fills against several makers. Size distributions differ between raw and aggregated tapes.
• Venue and instrument. Spot and perpetual tapes for the same asset are genuinely different. Multi-venue aggregation produces a fifth picture again.

None of this makes footprint unreliable; it makes it specified. Before comparing two footprints, ask which tape, which binning and which classification. That is the same hygiene that applies to order flow imbalance generally, and it is why the engine that draws a footprint should be explicit about its own construction rather than presenting the output as a universal fact.

How footprint is used on The Confluence Show

The Confluence Show is a live, AI-run market analysis broadcast. Its analyst, NAIRO, reads raw trades, order books and positioning 24 hours a day through the Confluence Engine — more than forty analytical layers computed in-house — and footprint is one of the layers that feeds several others rather than being an end in itself.

Concretely: per-bar bid/ask profiles are reconstructed from stored trades, which is why the layer exists at all — no footprint can be rebuilt from candles. Those profiles then supply the delta that gates absorption detection, the diagonal ratios that produce stacked imbalances and conflict zones, the POC that drives POC migration and value-area work, and the per-bar delta that accumulates into CVD. Every reading is stamped with the moment it became knowable, so nothing is drawn earlier than it could have been seen.

In the live room you can toggle these layers yourself and inspect the bar under discussion while NAIRO is talking about it. How it works explains the engine; watch the show free on a delay if you would rather see it read out loud first. It is educational market analysis, not financial advice, and it issues no signals.