Order-flow confirmation at a zone: write it so it can be scored

It is 09:33 ET on Tuesday 22 September 2026 (20:33 ICT). NQ trades 30,868.50, above Monday’s regular-session high of 30,862.75, with positive delta since the open. It “looks confirmed”. At 10:25 ET on Monday 2 November (22:25 ICT), NQ crosses Friday’s high with positive delta and looks confirmed again. The first trade ends the day 105 points up; the second is stopped out three minutes after entry.

Same feeling, opposite endings. This lesson replaces the feeling with a rule written before the open, applied the same way on both days, and logged every time so it can one day be scored.

Key terms

Order-flow confirmation (xác nhận order flow): a written, checkable condition on footprint, delta or DOM data, observed at a zone within a fixed time window, that must be true before you enter Definition. If two traders get different answers from the same chart, it is not a rule yet.

Zone (vùng): a price or narrow range from your map (M4.07) where you decided before the open to watch order flow.

Trigger (điểm kích hoạt) and verdict (phán quyết): the trigger is the moment your rule allows an entry. The verdict is M4.01’s judgement on whether the auction accepted the level, and it arrives later.

Scorecard column (cột chấm điểm): what you log on every touch of a zone, traded or not.

Terms you need. Delta is buyer-initiated minus seller-initiated contracts in a period (M3.04). Delta share is that delta divided by the period’s volume: +2,865 on 63,277 contracts is +4.5%. ARR is the average 09:30–16:00 ET range of the last 10 sessions, a ruler standing in for ATR.

Chart first: one probe, one written rule

How it works

Why confirmation has to be written down

Start with what the zone gives you. Our lab tested pre-market levels against random prices at the same distance: on 721 sessions of the Nasdaq-100 and S&P 500 ETFs, every family was touched and respected at the placebo rate Measured: no edgeNasdaq index. On NQ and ES futures themselves (5-minute bars, 46 sessions, autumn 2026), prior-day highs, lows and closes were respected less often than random prices, 23% against 35% Measured: no edgeNQ. A zone tells you where to look, not the odds.

So any edge has to come from what you require at the zone, and order flow describes the interval it was measured in. Order-flow imbalance explains most of the same interval’s price change and forecasts little after it, measured on NYSE stocks MeasuredOther market; on NQ, one practitioner’s study found 1-minute delta explains 43.6% of the same bar’s move and almost none of the next PlausibleOne trader's study. Positive delta during a break says buyers led that half hour. Whether requiring it improves the next hour is not yet measured, and a feeling can never be scored. Hence a written rule.

The six fields

Every confirmation rule in this course has the same six fields. Here they are filled in for R1, the breakout rule used in this lesson House rule:

• Field: 1. Zone and family — What you write: which zone, which side, which entry family (M6.01); R1: breakout above a prior-day high: prior regular-session high; breakout long.

• Field: 2. Observation — What you write: the footprint, delta or DOM fact, with a number; R1: breakout above a prior-day high: delta share of the 30-minute period that closes beyond is at least +1%.

• Field: 3. Time window — What you write: when the check is read, and when it expires; R1: breakout above a prior-day high: only at 30-minute closes (:00 and :30 ET); the first or second close after the probe; then the zone is done for the day.

• Field: 4. Distance — What you write: how far from the zone the observation counts; R1: breakout above a prior-day high: the close is at least 2 points beyond the high and at most 0.25 × ARR beyond (no chasing).

• Field: 5. What rejects it — What you write: what cancels the setup before entry, and what ends the trade after; R1: breakout above a prior-day high: a 30-minute close back inside before the trigger cancels it; after entry, a stop 0.075 × ARR inside the high, a 30-minute close back inside, or flat at 16:00 ET.

• Field: 6. Scorecard column — What you write: what you log for every touch; R1: breakout above a prior-day high: flow_confirmed (Y/N), trigger time, R, and the M4.01 verdict at the second close.

The thresholds are round conventions, not tuned values; once written they stay fixed, because changing one after seeing results fits the rule to the past (M10.04). The 0.075 × ARR stop borrows the “traded through” distance our lab uses to score levels.

Joining M4.01 and the Desk plan

M4.01 asks for a full verdict: at least 30 minutes beyond, volume beyond at 1.5× the usual for that hour, and two consecutive 30-minute closes. That takes about an hour. The Desk plan (DESK2.05) acts at a single 30-minute close. R1 gives each a job. The trigger is the Desk plan’s 30-minute close plus the flow check: early, and weaker. The verdict from M4.01 is logged at the second close as its own column and does not change the R1 trade. After enough rows you can compare R1 alone with R1 kept only when M4.01 says “accepted”, on the same touches: speed against certainty becomes measurable.

One rule per entry family

The same six fields fit the other three families from M6.01. These are starting templates House rule:

• Family: Rejection (M6.04) — Observation: probe’s period closes back inside, delta share −1% or lower (probe above); Window: first close after the probe; Distance: probe extreme ≤ 0.15 × ARR beyond; Rejected by: next close back beyond.

• Family: Retest (M6.03) — Observation: after an R1 trigger, a 5-minute bar at the edge closes away, delta share +1% or more; Window: 90 min after the trigger; Distance: bar low ≤ 0.05 × ARR from the edge; Rejected by: a 30-minute close inside.

• Family: Continuation (M6.05) — Observation: after a 15-minute pause beyond the zone, a 5-minute bar breaks the pause high, delta share +1% or more; Window: until 12:00 ET; Distance: pause 0.10–0.30 × ARR beyond; Rejected by: a 30-minute close inside.

Why it is still worth knowing

A written rule gives you vocabulary (“probe, no trigger, chase”) instead of “it looked strong”, describes the auction at the moment of decision, places risk before entry (field 5 is the stop), and turns the most repeated idea in order-flow trading into a hypothesis you can test on your own data.

What the data says

Worked example: 22 September 2026, long side

Real NQ data, December 2026 contract, one MNQ ($0.50 a tick). This is one day, chosen to teach the steps, not evidence for the rule.

1. Before the open. Zone: Monday’s regular-session high, 30,862.75. ARR = 317.2 points, so the distance window is 2 to 79.30 points beyond, and the stop sits 0.075 × 317.2 = 23.79 points inside: 30,839.00 after rounding to the tick.
2. 09:33 ET (20:33 ICT): probe at 30,868.50. No action. Nothing is read until a 30-minute close.
3. 10:00 ET: first close at 30,924.25, 61.50 points beyond (pass). The period traded 63,277 contracts with delta +2,865: share +4.5% (pass). First close after the probe (pass). Trigger. Long 1 MNQ at 30,924.25.
4. Risk. 30,924.25 − 30,839.00 = 85.25 points = 341 ticks = $170.50 per MNQ ($1,705 on one NQ), before costs.
5. 10:30 ET: M4.01 verdict, logged. Time: 58 of 60 minutes beyond (pass). Closes: 10:00 and 10:30 beyond (pass). Volume beyond: 89,744 contracts against 88,248 usual for that hour over the prior five sessions, 1.02× against 1.5× (fail). Not accepted; under R1 that is a logged column, so the trade stays on.
6. Exit. No 30-minute close came back inside and the stop was never touched. The high was 31,065.50 at 15:53; flat at 16:00 at 31,029.25: +105.00 points, +1.23R, $210 per MNQ before costs.

The scorecard row:

The mirror and the failure

Short side. R1 mirrored: the zone is the prior-day low, the close must sit 2 points to 0.25 × ARR below it, the closing period’s delta share must be −1% or lower, and the stop goes 0.075 × ARR above the low. Wednesday 23 September: zone 30,763.50 (ARR 321.3), probe 10:10 ET (21:10 ICT). The 10:30 close was 30,684.50, 79.00 points below, inside the 80.33-point limit; delta share −2.9%. Short 1 MNQ at 30,684.50, stop 30,787.50, risk 103.00 points ($206). The stop traded at 10:54 ET: −1R.

The failure. Monday 2 November 2026, zone Friday’s high 28,860.50 (ARR 401.9). Probe 10:25 ET (22:25 ICT). The 10:30 close was 28,887.25, 26.75 points beyond; delta +558 on 51,248 contracts, a share of +1.1%. Every field passed. Long at 28,887.25, stop 28,830.25, risk 57.00 points ($114). Three minutes later NQ traded 28,830.00, one tick through the stop: −1R plus any slippage. By 16:00 NQ was more than 700 points under the zone. Nothing was done wrong: the rule was written, followed and logged, and it lost.

All five triggers. Across the 53 sessions in our lab’s NQ tick data from 25 August to 6 November 2026, the open sat inside the prior day’s range and price then probed the prior high or low 24 times. R1 triggered 5 times Definition:

• Date (2026): Tue 22 Sep — Side: long; Entry: 30,924.25; Exit: 31,029.25, flat 16:00; R: +1.23.

• Date (2026): Wed 23 Sep — Side: short; Entry: 30,684.50; Exit: 30,787.50, stop; R: −1.00.

• Date (2026): Fri 25 Sep — Side: long; Entry: 30,883.25; Exit: 30,801.50, stop; R: −1.00.

• Date (2026): Wed 30 Sep — Side: long; Entry: 30,799.75; Exit: 30,716.75, close back inside at 16:00; R: −0.83.

• Date (2026): Mon 2 Nov — Side: long; Entry: 28,887.25; Exit: 28,830.25, stop; R: −1.00.

Total −2.60R before costs. Of the other 19 probes, 11 closed back inside first, 5 were too far beyond (chase) and 3 failed the flow check. This describes five events; it is not a test. Our lab estimates that a 10-point difference in hold rate needs about 150 to 200 touches to stand out from noise.

Do it on the platform

Prop-firm note. R1’s stops here ran 57 to 103 NQ points: $1,140 to $2,060 on one NQ, which can exceed an evaluation account’s daily loss limit, so this lesson uses one MNQ. Evaluation accounts fill stops in a simulator; the one-tick stop touch on 2 November could fill worse live. The Prop Firm course quotes the firm’s dated rules.

Common mistakes

Common mistake: Treating a passed confirmation as a forecast. The flow check describes the half hour that just closed. Whether requiring it improves the next hour is not yet measured; 22 September and 2 November passed the same fields and ended in opposite ways.

Common mistake: Reading five rows as the rule’s edge. −2.60R over five triggers says nothing about R1, just as five winners would not have proved it; a base rate from a handful of rows is noise.

Common mistake: Editing the threshold after a loss. Moving the delta share from 1% to 1.2% because one trade passed at 1.1% fits the rule to that trade. Write a second rule, R1b, and log both on every touch instead.

Drill: apply R1 to three probes

For each probe, write “trigger”, “no trigger” or “rejected”, and the R1 field that decided it.

1. Mon 28 Sep 2026, short. Zone: prior-day low 30,684.00; ARR 356.6. Probe 09:31 ET. 10:00 close 30,649.00, delta −505 on 68,055 contracts. 10:30 close 30,544.00, delta −676 on 48,888.
2. Thu 1 Oct 2026, short. Zone: prior-day low 30,680.75; ARR 323.5. Probe 10:01 ET. 10:30 close 30,630.25, delta −772 on 82,104. 11:00 close 30,618.75, delta +786 on 60,222.
3. Mon 26 Oct 2026, long. Zone: prior-day high 29,326.00; ARR 444.6. Probe 09:39 ET. 10:00 close 29,205.75.

What we don’t know yet

• Whether order-flow confirmation at a zone beats the zone alone, as R or as hold rate, on NQ: not yet measured. It is the first forward test on our lab’s list.
• Whether R1’s 30-minute close does better or worse than a faster 5-minute version: not yet measured.
• Whether M4.01’s 1.5× volume check suits NQ: it passed on none of the five R1 triggers. Five rows cannot tell; log it.
• Whether delta at a zone matters more in some volatility or gamma regimes: not yet measured (OPF9.01 frames it as a hypothesis).

Recap

• A confirmation is a written rule with six fields: zone and family, observation, window, distance, what rejects it, and the scorecard column.
• Order flow describes the interval it was measured in; whether requiring it at a zone improves results is not yet measured.
• R1 triggers at a 30-minute close with a delta check; M4.01’s verdict is logged at the second close, so the two can be compared later.
• Followed rules still lose: 22 September and 2 November passed the same fields.
• Log every touch, traded or not; never edit a threshold because of one result.

Further reading

• Order Flow, Measured: All claims. A public, pre-registered practitioner study on NQ, 2018–2026: claims FP-02 and DA-02, and the Elements & Confluences pages.
• Cont, Kukanov and Stoikov (2014), “The price impact of order book events”, Journal of Financial Econometrics 12(1). Order-flow imbalance and same-interval price change, on US stocks.
• Dalton, Jones and Dalton, Mind Over Markets (2nd ed., 2013). Time, volume and the 30-minute period as the vocabulary of acceptance.
• Our lab: the pre-market level study (721 sessions plus a 46-session futures check) finds no level family beating random prices and names order-flow confirmation as the part it did not test. Every dated example here comes from our lab’s NQ tick recordings, 25 August to 6 November 2026.

Next: M5.01 opens the setups module with the failed breakout at a level. Every setup in M5 carries a confirmation rule in this six-field format, so you can log it from the first session.

Educational content only. Not financial advice. Futures trading involves substantial risk of loss.