What it is
A stacked imbalance is an order flow footprint pattern: several adjacent price levels in a row all print one-sided aggression. At each level, aggressive buyers lift offers far more than sellers hit bids, or the reverse. Instead of a single tilted cell, you get layer after layer of the same tilt stacked on top of each other. It describes one side taking liquidity across a contiguous run of prices, quickly and repeatedly.
The more levels in the run and the more consistent the direction, the more concentrated the pushing force was during that stretch. It is a description of aggression that has already happened, not a statement about what comes next.
Keep it separate from depth-of-market imbalance. The imbalance here is trade-based aggression at each price in the footprint, not resting bid and ask size sitting in the book. Those are two different measurements: executed aggression versus unfilled orders.
What it looks like in order-by-order data
With L3 order-by-order data, every print carries an aggressor side and a size. Aggregate those prints by price and you get the footprint. A stacked imbalance shows up as neighbouring cells lighting up in the same direction, one after another, each with a clearly dominant side.
It tends to appear during fast one-way moves, and usually travels with a lopsided aggressive-buy share and a heavy net aggressive delta. A quick read:
- adjacent price levels, same direction, back to back
- each cell clearly one-sided rather than marginally tilted
- price travelling quickly in one direction rather than chopping
A single tilted cell is just an ordinary imbalance. The word stacked is about the run.
Common misconceptions and limits
The most common mistake is treating book imbalance and footprint imbalance as the same thing. One measures resting orders across the top ten levels; the other measures executed aggression at each price. A finding about one does not transfer to the other.
The second mistake is reading a stack as a guarantee. A stack only says that one side was aggressively taking liquidity a moment ago. It can still be absorbed, and it can be interrupted by the other side stepping in.
Other limits worth keeping in mind: a stack is a finished print, so it can be worked off as the session develops; iceberg orders can only be inferred from repeated refills, never confirmed; and different tools use different thresholds for what counts as an imbalance, so the same tape may or may not qualify depending on the definition you are running.
What the GC research says
Scope matters here: this is COMEX gold futures (GC), TradeWhy research, data through 2026-08-31. These findings hold only within that scope and should not be extended beyond it.
Within that scope, the relevant entry tested using ten-level book imbalance as a primary signal. It failed three independent checks. It qualifies only as a weak secondary input, worth roughly 8 percentage points of marginal improvement, and it is graded as disproven — the approach did not pass testing on the GC data.
Be precise about what that covers. The entry is about book, or depth, imbalance, not per-price trade imbalance in the footprint. Do not move that result onto stacked imbalances, and do not read it as evidence either way about them. Stacked imbalance itself was not separately tested in this body of work, so treat it as a descriptive pattern rather than a validated signal.
How to see it in TradeWhy
TradeWhy is an AI market-intent analysis and strategy-validation tool for CME Group futures day traders. The first release covers COMEX gold (GC) and CME Nasdaq (NQ). It combines real-time macro information, price response and L3 order-by-order behaviour to explain what the dominant force is currently doing.
Internally, TradeWhy's engine measures the aggressive-buy share over the last 60 seconds (0-100, with 50 as balance) and one-minute net aggressive volume (buys minus sells, in contracts) to gauge the direction and intensity of aggression. These are engine-internal readings, not published scores and not signals, and they say nothing about where price goes next.
Repeated behaviour can be organised into strategies, validated historically and forward, and then added to a strategy portfolio. TradeWhy explains market behaviour and helps users research and validate strategies. It does not give buy or sell advice or guarantee returns, and the final trading decision is the user's.
FAQ
What does stacked imbalance mean?
It means a run of adjacent price levels in the footprint where aggression is clearly one-sided at every level, so the tilt stacks up instead of appearing at one price. It shows one side taking liquidity across several ticks in a row.
Is a stacked imbalance the same as a book imbalance?
No. Book imbalance compares resting size across the top levels of the book, while a stacked imbalance is about executed aggression at each price in the footprint. Research in this pack tested book imbalance, so that result should not be reused for footprint stacks.
Does a stacked imbalance mean price has to continue in that direction?
No. A stack describes aggression that already happened. It can be absorbed or interrupted, and it does not predict price. TradeWhy does not give buy or sell advice or guarantee returns.
Do more stacked levels mean a more reliable signal?
More levels only mean the push was more concentrated during that stretch, not that the pattern is more predictive. The related approach tested in this research did not pass testing on the GC data, so level count should not be used as a reliability scale.
Can I see a stacked imbalance reading in TradeWhy?
TradeWhy's product statements do not list stacked imbalance as a separate reading. Its engine internally measures the aggressive-buy share and one-minute net aggressive volume to describe the direction and intensity of aggression; these are engine-internal readings, not displayed signals.
Written by AI from TradeWhy's metric definitions, research findings and published product statements, then rule-checked. Research figures come from TradeWhy's study of COMEX gold futures (GC), data through 2026-08-31, and apply to GC only. Analysis only, not investment advice.