Skip to content

Stacked imbalances: initiation or climax? Where the stack sits decides the reading

By Author of ORDER FLOW — Market Mechanics

Read four explanations of stacked imbalances and you get four instructions. One indicator's documentation says a stack means the move is likely to continue. A forum guide warns that stacks usually continue and should not be faded without more evidence. A strategy write-up does the opposite and fades a stack at the session high when the next candle fails to extend. A platform's guide says the stack itself is not the signal; the retest is.

They are not all wrong. Each describes some stacks correctly, because the same mark appears at the start of a move and at its end.

What a stack records, and what it leaves out

A stacked imbalance is three or more consecutive price levels where aggressive volume on one side beat the other side's aggressive volume on the diagonal by at least the ratio you set: the ask at each level against the bid one tick below. Both numbers are settings. A common reference is 3:1 over three levels; the book's own stacked-imbalance setup asks for 4:1. It records urgency: someone paid the spread level after level instead of waiting.

What it leaves out is where in the move that urgency happened. An initiation stack, in the book's terms, prints when price breaks out of a range or through a level on the map, after a period of compression. A climax stack prints after a leg that has already travelled a long way with no pause, when the last buyers arrive at once, and its cells can match an initiation stack exactly.

The reason location matters is mechanical. After an initiation stack, the participants caught on the wrong side of the broken level have not closed yet; when they do, they close with market orders in the direction of the stack, and that adds fuel. After a long leg with no pause, most of the participants who were going to buy already have. A climax stack is what it looks like when the remaining ones arrive at once and the passive orders on the other side fill them instead of being swept. With the same cells, the supply of future aggression is very different.

The same stack in two places 8 candles within 12.00 points A 18.00 pts from the origin B 52.50 pts from the origin Cells in A and in B: 210 vs 40 · 190 vs 35 · 230 vs 45 Diagonal ratios: 5.3 · 5.4 · 5.1 Identical cells, with very different travel behind them.
Schematic of one sequence of 5-minute NQ candles with two buying stacks whose cells are identical. Values are illustrative.

In the schematic, stacks A and B have the same three cells and the same diagonal ratios, the lowest of them 5.1:1, so both pass the 3:1 test and the 4:1 test. A prints after 8 candles held within 12.00 points and sits 18.00 points above the low of that compression. B prints 52.50 points above the pullback low that started its leg, after 8 candles that covered 40.75 points between them. Any tool that flags stacks flags both.

The delta behind the stack

The second thing the book checks is how the stack candle's delta compares with the pushes that came before it in the same impulse, and what the cumulative delta was doing beforehand. An initiation stack tends to follow a CVD that was already rising while price went nowhere: aggression being absorbed inside the range until it stops being absorbed. A climax stack tends to be the largest delta of a leg that was already one-sided.

Delta per candle and CVD on the same record Compression: CVD +520 with price +5.25 CVD Delta per candle A: delta +620 first push of the impulse B: delta +980 B: delta +980, 2.7× the largest of 9 earlier pushes (+360) Volume vs the mean of the previous 10 candles: 250% in A and 250% in B Only the delta history separates them; relative volume is the same.
Schematic of the delta of each candle and the cumulative delta for the same sequence. Values are illustrative.

During the compression, the CVD rises +520 while price gains 5.25 points. Stack A's candle, with a delta of +620, is the first push of a new impulse. Stack B's candle prints +980 after 9 earlier pushes in the same leg, 2.7 times the largest of them. Both candles traded 250% of the average volume of the ten candles before them, so a relative-volume filter accepts both.

What the retest shows

Location and delta history are known when the stack prints, but they only shift the odds. The book does not enter on the stack candle. It waits for price to come back to the stack zone and reads what the return looks like. A pullback on light volume with negative delta that does not close below the lowest stacked level is a test the zone passes. A return on volume equal to or above the stack candle's, with strong delta against it, is a counterattack.

Return to A stack zone 3,825 2,600 2,100 1,600 Returns on 1,600 contracts, 42% of stack A's candle Delta −190, close 21,319.75, above the zone Next candle: delta +210 Return to B stack zone 6,400 6,600 4,100 Returns on 6,600 contracts, 103% of stack B's candle Delta −1,050, close 21,364.00, below the zone The counterattack matches the stack The signal is what happens when price comes back to the stack.
Schematic of the first return to each stack zone, with the volume of every candle. Values are illustrative.

Price comes back to A on 1,600 contracts, 42% of the stack candle, with a delta of −190. It trades below the zone, closes at 21,319.75, above it, and the next candle prints +210. Price comes back to B on 6,600 contracts, 103% of the stack candle, with a delta of −1,050, and closes at 21,364.00, below the zone. This is the same logic that separates a breakout that starts a trend from one that ends a probe: what follows the candle decides.

Limitations and false positives

A thin book prints stacks for free. When resting liquidity is sparse, price crosses several levels on a handful of contracts and every one of them qualifies. On NQ this happens often enough that a stack has to be checked against the volume behind it before anything else.

“Largest delta of the leg” depends on the candle interval. Change the bar length and the same flow is grouped differently, so a different candle can carry the largest delta. The comparison is only meaningful on a fixed interval decided in advance.

Distance has no universal threshold. How far is “extended” depends on the volatility regime. The book gives a working figure for NQ and says it must be recalibrated; a fixed number carried across regimes will misclassify stacks.

A climax stack lowers the odds. It makes continuation less likely without saying that price will turn. Trends extend through stacks that look climactic, and fading them on that basis alone is its own mistake.

The retest may never come. Price often leaves the strongest initiation stacks behind without returning. Waiting for the retest means missing some of those moves.

A stack zone expires. It works as a reference until price comes back to it on above-average volume and trades through without absorption. After that, the liquidity that created it has been consumed and the zone no longer marks anything.

Data releases. In the seconds after a scheduled release the book empties and refills, and almost every level prints as an imbalance. None of the three readings applies until the book is dense again.