A footprint chart is only as useful as the aggregation underneath it. This page documents how ATAS builds footprint bars, which parameters control the numbers inside each price row, and which settings change the interpretation of a cluster rather than merely its appearance.
01 Aggregation families and bar construction
ATAS accepts three aggregation families. Clock intervals divide the session by time. Tick counts divide it by traded events. Volume and range bars divide it by activity or by price travel. Each family produces bars with different internal structure, and the footprint row array inherits that structure directly.
| Family | Written as | Effect on cluster rows |
|---|---|---|
| Clock interval | 5m | Rows accumulate until the clock closes the bar |
| Tick count | 1500 ticks | Rows fill evenly; bar closes on the tick that reaches the count |
| Range | 8R | Bar closes when price travels the configured range |
| Volume | 25000 V | Bar closes when traded volume reaches the threshold |
A tick-count aggregation is usually the first choice when the goal is comparable density across instruments, because the number of price rows per bar stays roughly stable while a clock bar can be almost empty in a quiet period. The trade-off is that timestamps no longer line up, which matters on the replay and data feeds page.
02 Cluster modes and what each one hides
Four cluster modes cover most desks. Bid x Ask keeps both sides per row. Volume collapses them. Delta prints the signed difference. Diagonal imbalance prints a marker only where the configured ratio is exceeded.
- Bid x Ask keeps the aggressor attribution intact, at the cost of horizontal space: two numbers, a separator and a colour rule per row.
- Volume is the most compact and the least informative about direction.
- Delta is a derived quantity, so it cannot be read back into two sides without the underlying cluster.
- Diagonal imbalance is a filter, not a measurement; changing the ratio changes how many rows appear to exist.
The imbalance ratio is a multiplier against the diagonal neighbour, not an absolute size. A filter tuned at 3.0 on one contract therefore stays meaningful on a contract whose typical print size is several times larger.
03 Row height, tick size and text fit
Row height is derived from instrument tick size and a per-panel multiplier. Two panels showing the same instrument with different multipliers are still consistent in value, only compressed or expanded in pixels. When a panel is scrolled to a narrow price range, the text inside a row can collide; the usual remedy is either a reduction of the row multiplier or a switch to a single-number cluster mode.
- row 1.0x — one tick, one row; the default mapping.
- row 2.0x — halves the number of rows for the same price travel.
- auto — expands until the row can hold the widest number in the bar.
04 Summary rows and per-bar totals
Each footprint bar can carry a summary row: total volume, total delta, or the largest single print. The summary row is a separate layer drawn above or below the price rows, so it survives changes to the cluster mode. On a narrow bar the summary row is often the only readable element, which makes it worth configured deliberately rather than left at a default.
Profiles attached to the same bar are covered on the cluster chart and TPO page, and the ladder that consumes the same prints is described in Smart DOM depth settings.
05 Filters, colour rules and reproducibility
Footprint colouring follows value bands: a row is tinted according to how its size compares with a reference value, and the reference can be local to the bar or global to the window. Local references make every bar look busy; global references make quiet bars look empty. Both are legitimate, but the choice should be recorded, because a screenshot without it cannot be reproduced.
Filters operate before rendering. A minimum size filter, for example, suppresses rows below a threshold and changes the visual density without changing the underlying series. When a footprint chart looks different from a colleague's, the filter threshold is the first thing to compare, followed by the cluster mode and then the aggregation family.
