Footprint imbalance
Also called: bid/ask imbalance · diagonal imbalance · delta imbalance
A footprint imbalance compares two cells diagonally — buying at one price against selling at the price one tick below — and flags the cell when one side exceeds the other by a set ratio, usually three to one. Without an absolute volume floor, the flag is meaningless.
Why the comparison is diagonal
The comparison looks wrong until you think about the spread. When the market is bid at 5 318.00 and offered at 5 318.25, an aggressive buyer trades at 5 318.25 and an aggressive seller trades at 5 318.00 — at the same instant, against the same book. The diagonal pairs two populations that were competing.
Comparing ask and bid volume at the same price pairs events that could not have happened simultaneously: the quote had to move in between. So a buy imbalance is ask volume at price P over bid volume one tick lower, and a sell imbalance is the mirror. Everything else on a footprint chart is arithmetic on those two numbers.
The ratio is a setting; the floor matters more
Three to one is the default in most platforms. It has no published derivation and no public test — it propagated because it shipped as a default, and 2:1 or 4:1 are equally defensible. Treat it as a knob you fixed and wrote down, not as a property of the market.
The floor is the setting nobody configures and the one that decides whether the flag means anything. A cell holding 12 contracts against 3 clears 4:1 and carries fifteen contracts of information. Set a minimum absolute volume per cell, scaled to the instrument and the bar duration, and most of the noise disappears before you look.
One imbalance is not a signal
A single flagged cell says aggression arrived unevenly at one price during one bar. That happens constantly. The readings that carry structure are runs of three or more on the same side — see stacked imbalances — or a flag sitting on a level you had marked before the bar printed. Alone, imbalance highlighting is a colouring algorithm applied to noise.
A worked example
In a synthetic ES sequence, one bar prints 412 contracts at the ask at 5 318.25 against 96 at the bid at 5 318.00 — just over four to one, on 508 contracts.
Ten ticks higher, the same bar prints 14 at the ask at 5 320.75 against 3 at the bid at 5 320.50. The ratio is nearly five to one, larger than the first, on seventeen contracts. With no floor configured the platform paints both cells identically, and the weaker one sits at the bar extreme where your eye goes first.
The trap
Ratios explode where denominators are small, and denominators are smallest at the edges of a bar, where price spent the least time. The flags cluster exactly where the sample is thinnest — which is also where the eye lands. The display concentrates your attention on its own least reliable output.
The second half is the setting itself. Lower the threshold to 2:1 and the chart fills with flags; raise it to 4:1 and half of them vanish, on identical data. Before building anything on an imbalance, move the ratio one step and move the floor. What survives both is a reading. What does not was a rendering choice.
Frequently asked
- What ratio should I use for footprint imbalances?
- There is no derived answer, and anyone giving you one with confidence is repeating a platform default. Three to one is the common convention. What matters more is pairing it with an absolute minimum volume per cell and never changing either mid-session, so that two charts you compare were built the same way.
- Why is the comparison diagonal and not at the same price?
- Because the bid at one price and the ask at the price above are the two sides of a one-tick spread, quoted at the same moment. Buyers and sellers hitting that spread faced the same market. Comparing bid and ask at a single price compares events separated by a quote move.
- Does an imbalance mean a large trader was active?
- No. It means aggression at one price outweighed aggression at the neighbouring price by your chosen ratio — one large participant, many small ones, or an algorithm slicing an order. Footprint data aggregates by price and carries no order identity; that needs market-by-order data.