Stacked-imbalance continuation

It is 10:20 ET (22:20 ICT, winter). NQ has driven from 18,040 to 18,059 since 09:50. The 5-minute bar that closed at 10:15 shows four consecutive buying imbalances on its footprint, and the bar that just closed went nowhere: a small inside bar at the top of the move. Many order-flow courses call the stack an institutional campaign and the pause a gift: join it, because campaigns do not stop after one quiet bar. This lesson turns the moment into a written rule you can score, and says what the data shows: on NQ the ingredients showed no edge, and the full rule has not been tested.

Key term

Stacked-imbalance continuation (tiếp diễn với stacked imbalance): a setup that joins a directional drive after a bar prints a stacked imbalance in the drive’s direction, entering on the first pause rather than inside the stacked bar Definition.

Term you need: a stacked imbalance is three or more consecutive prices in one bar where the aggressive side outweighs the opposite side on the diagonal by a set ratio, for example 300% (M3.03). The ratio and the count are settings, not tested values. The stack’s origin is its first level: the lowest stacked price in a buying stack, the highest in a selling stack.

The anatomy of the rule

Every setup in this module has the same five parts: context → trigger → invalidation → target → R:R.

• Context: a drive already under way, read as a description: price has travelled one way for several bars. Whether it continues is the open question. The map from before the open (M4.07) tells you what references sit ahead House rule.
• Trigger: a bar in the drive’s direction prints a stack of at least three imbalances at your chosen ratio, and the next bar pauses: an inside bar, or a bar that holds above the stack’s top with little net delta House rule. Entering inside the stacked bar buys the most aggressive prices of the move; the pause gives a defined place to be wrong. That is a risk choice, not a measured improvement.
• Invalidation: a hard stop one tick beyond the origin, placed with the entry House rule. Price trading back through the whole stack means the drive you joined has been fully retraced; the reason for the trade is gone.
• Target: the nearest reference ahead that you mapped before the open, fixed before the entry, plus a time stop. Mapped levels were respected no more often than random prices in our lab’s placebo study Measured: no edgeNasdaq index (M5.01, second box), so the target is a place to take profit, not a forecast of where the move ends.
• R:R and break-even: stop and target together imply the win rate you need after costs. Compute it before the click (M6.08). If nothing tells you the rule beats that rate, the honest status is unknown.

Myth / hypothesis: “A stack is an institution committing to a campaign; campaigns do not end on one pause, and the stack’s prices become the floor they defend.” Lore The footprint shows aggressive volume at prices, not who traded or why. The testable parts were tested on NQ and not supported (box below).

Myth / hypothesis: “Stacks with expanding delta are initiative and continue; stacks with flat delta into a level are absorption in disguise.” Lore In the same public study, initiative activity did not continue more often than baseline, extreme delta did not forecast the next 15 minutes, and absorption did not predict reversals. Neither delta reading has been shown to sort stacks into winners and losers.

Myth / hypothesis: “Failed stacks do not drift, they snap: the stacked buyers are trapped, and their exits drive a sharp reversal.” Lore The closest public tests were inconclusive: heavy buying at a high that fails to extend, then reverses (+1.9 percentage points, 95% CI −0.4 to +4.3, n = 11,846), and a failed breakout on heavy aggression accelerating the other way (+1.7 pp, CI −0.1 to +3.5, n = 14,508). A failed stack is a reason to be out, not a reason to reverse.

Why it is still worth knowing

• Shared vocabulary. “A four-level buying stack in the 10:10 bar, then an inside bar” describes the auction without guessing intent.
• Framing risk. The origin is an objective, nearby price where the trade is plainly wrong, which keeps each attempt small.
• A testable rule. A written trigger, stop, target and time stop can be scored against a baseline (see the drill): the only route from “unproven” to “measured”.

What the data says

The promise is that joining a stack on the pause leads to continuation. This exact rule is not yet measured on NQ; its ingredients were tested in a public practitioner study, and none showed an edge.

A second result bears on the target. Courses often say a stack starts a move, so give the target room. In the same study, footprint signals widened the best and the worst excursion together: a volatility flag, not a direction flag PlausibleOne trader's study.

So the honest status: a rule-based continuation entry with a pre-set stop, whose edge is unproven. You can trade it small, as a defined-risk rule you are scoring. You cannot call it high odds.

Worked example: the 10:10 stack

Synthetic session, built for this lesson, not a real date. NQ, 5-minute footprint with rows of 1 point (4 ticks per row), $5 a tick and $20 a point per contract; MNQ is one tenth of that.

• Map, written before the open. The overnight high at 18,072.00 is the nearest reference above. Scenario: if a buying stack forms in a drive and the next bar pauses, buy the pause with a stop one tick under the origin and a target at 18,072.00; no trade if the stop distance makes the arithmetic fail.
• 09:50–10:10 ET, the drive. NQ travels from 18,040 to 18,057 in four bars. Description only: buyers have been the aggressive side.
• 10:10–10:15 ET, the stack. The bar’s footprint shows buying imbalances at 18,054, 18,055, 18,056 and 18,057: asks of 410, 450, 480 and 420 against diagonal bids of 130, 135, 140 and 120, ratios of 3.2× to 3.5×. Bar delta +1,565 on 4,095 contracts; close 18,058.50. The origin is the 18,054 row.
• 10:15–10:20 ET, the pause. Inside bar, 18,056.50 to 18,059.00, delta +40, close 18,058.00. Trigger met.
• Plan. Long 1 NQ at 18,058.00 at the pause bar’s close. Stop 18,053.75, one tick under the origin row: 4.25 points = $85 ($8.50 on 1 MNQ). Target 18,072.00: 14 points = $280. Time stop: flat by 11:20 ET if neither has traded.
• Break-even. With an assumed $15 round turn (commission, fees and a tick of slippage), the trade nets +$265 on a win and −$100 on a loss, so it needs to win 100 ÷ 365 ≈ 27% of the time. How often this rule reaches its target on NQ is not yet measured, so you do not know whether it clears 27%.
• Outcome on this synthetic day. Price leaves the pause, prints a second, smaller stack at 18,063 at 10:30 and trades 18,072.00 at 10:55: +$265 after costs. One day is one outcome, not evidence; the figure shows the same entry stopped out.

The mirror and the failure

The short side (mirror). In a down-drive, a bar prints three or more selling imbalances (the bid at each price several times the ask one tick higher), and the next bar pauses near its low. Short the pause, stop one tick above the origin (the highest stacked price), target the nearest mapped reference below, same time stop. Flip the figure to see it. The evidence is the same: the public tests covered buying and selling imbalances.

The failure. The figure shows it. After the long at 18,058.00, price lifts to 18,058.75, then returns into the stack at 10:30 and trades through the origin by 10:40. The stop at 18,053.75 fills: −$100 after costs. The read was correct at 10:20; the drive ended. That is a normal outcome of this rule, not a broken rule. Do not reverse to short because “the stacked buyers are trapped”: that story was inconclusive in the public tests.

Prop-firm note. In a prop evaluation the stop is a real order placed with the entry. Evaluation accounts usually fill in simulation, and simulated fills are not live fills; tuning very tight brackets or automatic breakeven moves to win on simulated fills, or placing orders outside the best bid and offer, is treated as prohibited conduct by firms such as Topstep. A stop one tick beyond a 4-point stack is set by the structure, not by the simulator. The Prop Firm course (PROP1.04) covers the rules; check your firm’s help center before you trade.

Common mistakes

Common mistake: Reading a strong-looking stack as an edge. A stack records that aggression happened in that bar. On NQ, single imbalances did not lead to continuation and stacks did not hold on a retest. How strong a print looks and whether a signal has an edge are different questions.

Common mistake: Widening the target because “stacks start moves”. In the public study, footprint signals widened the worst excursion as much as the best, and no stop/target shape made them pay after costs. Set the target from the map and the break-even arithmetic, before the entry.

Drill: score twenty stack pauses against a baseline

Before you open a chart, write your rule: ratio (for example 300%), minimum stack count (3), what counts as a pause, stop one tick beyond the origin, target at the nearest mapped reference, time stop 60 minutes.

1. In NQ replay on a 5-minute footprint, find 20 stack-then-pause triggers in drives. Log time (ET), side, origin, entry, stop, target.
2. Score each: target first, stop first, or time stop, in R after costs.
3. Baseline: find 20 pauses after drive bars with no stack in the same sessions and score them with the same rule.
4. Compare both win rates with the break-even rate your stops and targets imply.

What we don’t know yet

• How often the stack-pause entry reaches a mapped target before its stop on NQ, against pauses with no stack: not yet measured.
• Whether entering on the pause beats entering inside the stacked bar, or not trading at all: not yet measured.
• One nearby result leans toward continuation: in the public study, aggressive volume surges at the end of a move reversed 2.8 percentage points less often than baseline (CI −5.0 to −0.7, n = 14,546). That is smaller than the study’s own 3-point cost line, and whether a stack adds anything to it is not yet measured.
• The open leads this module tracks are all graded Lore until a forward test, and none involves stacked imbalances.

Recap

• A stacked imbalance describes sustained aggression in one bar; it does not tell you who traded or what comes next.
• The rule: a stack in a drive, entry on the first pause, a hard stop one tick beyond the origin, a target from the map, a time stop, and the break-even win rate after costs, all written before the click.
• On NQ, single imbalances did not lead to continuation and stacks did not hold on a retest; the full rule is not yet measured, so its edge is unproven.
• A return through the origin is a normal ending: the stop handles it, and a trapped-trader story is no reason to reverse.

Quiz

Further reading

• Order Flow, Measured: All claims. A public, pre-registered practitioner study on NQ, 2018–2026. Read the imbalance rows (FP-06, FP-07, FP-08), the initiative row (DA-08), the trapped-buyer rows (DA-06, FP-10) and the page “Does a Wider Target Help?”.
• ATAS Knowledge Base: Stacked Imbalances. The platform’s own description of a diagonal stack indicator; set it up as in M3.03.
• Our lab: no NQ footprint test of its own yet; its pre-market level study found mapped levels respected at the placebo rate.

Prerequisites: M3.03 Imbalance & stacking, M4.01 Acceptance, and EV3.01 on sample size. Next: M5.10 Initial-balance breakout, another continuation rule, with the opening-range numbers that exist for it.

Educational content only, not financial advice. Futures trading involves substantial risk of loss. Synthetic examples are illustrations, not trade calls.