Overnight inventory

Monday 26 October 2026, 09:30 ET (20:30 ICT, summer time). NQ opens at 29,230.50. While you slept, the overnight session traded almost 197,000 contracts across a 330-point range, a quarter more than an average night, and most of that volume sat above Friday’s close. Is the open about to “correct” that long inventory, or is the overnight high the first thing to go?

You cannot know. What you can do is measure what the night left behind, in two minutes, and know which stories about it have been tested.

Key terms

The overnight session (ON; phiên qua đêm) is the Globex session from 18:00 ET to the 09:30 ET open (17:00 to 08:30 CT; 05:00 to 20:30 ICT in summer time) Definition. It trades fewer contracts per hour than RTH and reacts to Asian and European news and to the 08:30 ET releases.

Overnight inventory (tồn kho qua đêm) is the position the night built: if most overnight volume traded above the prior session’s close, the night is net long; mostly below, net short Definition. Alongside it you record:

• ON high and ON low: the night’s extremes.
• ON midpoint: (ON high + ON low) / 2, the middle of the night’s range. It is arithmetic, not a level anyone has to defend.
• ON volume vs its average: the night’s volume divided by the average of the last 20 nights. Above 1 means more positions changed hands than usual.

A poor high (đỉnh kém, M2.09) is an extreme printed by two or more periods at the same price, with no tail.

Chart first: one night’s inventory

How it works

Measuring the night in two minutes

Before the open, write five numbers on your sheet Definition:

1. ON high and ON low, with the time each printed.
2. ON midpoint.
3. ON volume / 20-night average.
4. Share of ON volume above the prior close (or simply: did the night spend most of its volume above or below it?).
5. Where the open sits: above or below the midpoint, inside or outside the night’s value area.

Then mark anything unfinished: a poor ON high or low (M2.09) and overnight single prints (M2.08). That is the whole read. It describes the night; it does not forecast the day.

Who is holding what at the open

The numbers tell you something real about positioning Context. On 26 October the open (29,230.50) sat above the midpoint and near the top of the night’s value, so most people who bought overnight were in profit and most who sold overnight were at a loss. On a night that traded 1.25 times its usual volume, that is a lot of open positions near the price.

What the numbers do not tell you is what those positions do next. Traders who are underwater can add, hold or cover; traders in profit can take it or add. The classic Market Profile literature turns this into two directional stories. Both have been tested on NQ.

Myth / hypothesis: “Overnight inventory gets corrected early: a net-long night is sold in the first part of RTH, a net-short night is bought.” Lore On NQ the measured effect leaned slightly against this story and did not survive the correction for testing many claims at once (numbers below). A long night is a description of who traded, not an instruction for the open.

Myth / hypothesis: “RTH goes back to finish what the night left ragged: a poor ON high or overnight single prints are the day’s first targets, and a failed test of the ON high or low reverses.” Lore The NQ tests found no edge for revisits of poor highs and lows, for single-print zones, or for reversals after failed tests of the overnight extremes. Our lab also found that the overnight high and low get touched and respected at the same rate as random prices the same distance from the open.

The ON high and low as references

The overnight extremes are still useful lines. They are where the night’s auction stopped, so the first RTH trade beyond one is a fair moment to ask the acceptance question from M2.07: does price build volume and time beyond the line (acceptance), or come straight back inside (rejection)? Definition That question describes what is happening; it does not make the line hold. Measured against random prices at the same distance, the ON high and low were not respected more often Measured: no edgeNasdaq index.

Why it is still worth knowing

• Shared vocabulary. Desk notes and AMT books talk about the overnight range, its midpoint and its inventory. You need to read them.
• Describing the auction. Opening inside the night’s value is a different morning from opening 100 points outside it, and the ON range tells you how much ground the night already covered.
• Framing risk and stops. The ON high and low are natural invalidation prices for an if-then plan, because a move through them changes the description of the morning.
• A hypothesis you can test. The correction and revisit stories are exactly the kind of claim you can log yourself, with a written rule (see the drill).

What the data says

Worked example: a heavy, net-long night (real data)

Monday 26 October 2026, NQ front month, from our lab’s tick recording.

The sheet at 09:25 ET (20:25 ICT):

• ON high 29,354.00 (printed at 08:26 ET, in one period: not poor). ON low 29,024.25 (04:11 ET). Range 329.75 points.
• Midpoint 29,189.13.
• ON volume 196,719 contracts, 1.25 times the average of the previous 20 nights: a heavy night.
• 70.5% of the night’s volume traded above Friday’s 16:00 ET close of 29,122.25: net long.
• The night’s value area roughly 29,050–29,250. Pre-open price about 29,231, near the top of it.

The watch-list, not a forecast:

• If RTH trades above 29,354 and builds time there, the night’s high is accepted and the long inventory is being extended. Invalid back below 29,354.
• If RTH trades below the midpoint and then the prior close (29,122.25), the long inventory is being unwound. Invalid back above the midpoint.
• Between the two, the night explains nothing yet.

What happened. The open printed 29,230.50. By 09:32 ET price had dropped to 29,165.25, below the midpoint: the “correction” story looked right for two minutes. It stopped 43 points above the prior close and rallied. At 10:18 ET NQ traded above the ON high; by 10:30 it was back at 29,295, below it again: rejected on the first look. From 10:35 it went back through, and after 10:40 it never traded below the ON high again in RTH. The RTH high was 29,572.00 at 11:02 ET, and the last trade before 16:00 was 29,439.75.

What the sheet gave you: two prices that changed the description of the morning, each with an invalidation. What it did not give you: which one would come first. The midpoint broke within two minutes and decided nothing; the first rejection at the ON high was undone within minutes.

The mirror and the failure

Mirror (net-short night). Flip every line: most overnight volume below the prior close, the open below the midpoint. The watch-list is the ON low (acceptance below means the short inventory is being extended; invalid back above it) and the midpoint and prior close above (reclaimed means the short inventory is being unwound; invalid back below the midpoint).

Failure: the poor high nobody went back to. Friday 21 August 2026, from the same recording. The ON high, 29,539.00, printed twice, at 07:36 and again at 08:45 ET, in two different periods and with no tail: a textbook poor high. 94.8% of the night’s volume was above the prior close of 29,306.50, and the open at 29,469.50 sat above the midpoint (29,406.63). The “unfinished business” story says the poor high comes first.

It did not. NQ fell below the midpoint at 09:31 ET, through the ON low (29,274.25) at 09:54 and down to 29,220.00 at 10:21. The RTH high was 29,488.75, at 11:53, 50 points short of the poor high, which stayed unvisited all session. On this day the “inventory correction” story would have looked right; on 26 October the “go finish the night’s high” story would have. Two stories, two days, each right once: that is what an untested rule looks like from the inside, and why the studies above count hundreds of sessions instead of remembering two.

Common mistake: Treating a base rate as an edge. “The ON high gets touched on most days” can be true and still be worthless: our lab found the ON high touched on 60% of sessions, and random prices at the same distance from the open on 59%. A touch rate only matters when you compare it with a placebo.

Common mistake: Weighting the levels by the night’s volume as if that were tested. Heavy overnight volume does mean more positions near the price, and it is worth recording. Whether a heavy night makes the ON high and low behave differently from a light night is not yet measured. Record it, log the outcome, and do not size up because the night “meant it”.

Common mistake: Treating the midpoint as the line that decides the day. It is the middle of a range, nothing more. On 26 October it broke in the first two minutes and the day finished 250 points above it.

News nights (CPI, payrolls, FOMC) need one extra line on the sheet. The overnight range and inventory are then built around a headline, and the open often gaps away from them. Our lab found that on mornings with a large gap the whole day is wider and every level, real or random, stalls less often PlausibleNasdaq index; FOMC days were too few to test. Before you trade an evaluation or funded account around those times, check the firm’s news-window rules (Prop Firm course).

Drill: log ten overnight sessions

For the next ten sessions, before 09:30 ET:

1. Write the five numbers from “Measuring the night” and mark any poor ON high or low.
2. Write a watch-list: the ON high and ON low with acceptance and rejection branches, and an invalidation for each.
3. After 11:00 ET, record what happened: which overnight extreme (if any) traded first, whether it was accepted or rejected, and whether a net-long night sold off first or a net-short night rallied first.

What we don’t know yet

• Whether ON volume relative to its average changes how the overnight high and low behave. Not yet measured.
• Whether poor overnight highs and lows (as opposed to all overnight extremes) are revisited more often than random prices on NQ futures. The NQ test of poor extremes in general was inconclusive; the overnight subset is not yet measured.
• Our lab’s placebo test used QQQ and SPY sessions with futures-based overnight levels; a full test on NQ futures sessions has not been run beyond the 46-session check.

Recap

• Measure the night before the open: ON high, ON low, midpoint, ON volume vs its 20-night average, and the share of volume above the prior close (net long or net short).
• These numbers describe who is holding what at the open; they are context, not a directional call.
• “Inventory gets corrected early” is Lore on NQ: the measured effect leaned slightly the other way (n = 1,818).
• The overnight high and low were touched and respected like random prices at the same distance (Measured: no edge); use them as acceptance questions and invalidation prices, with both branches written down.
• A poor overnight high is a line to watch, not a target: on 21 August 2026 it was never revisited.

Quiz

Further reading

• Market Profile (TPO), ATAS help: splitting overnight and RTH structure on the TPO chart.
• James Dalton and co-authors, Mind Over Markets and Markets in Profile: overnight inventory in the auction framework. Read the inventory chapters as description; the directional rules are the ones tested above.
• Following the Auction: Market Profile Basic Language: plain-English auction vocabulary.
• Order Flow, Measured: all claims: the public practitioner study behind OI-10, OI-11 and AP-05.
• Our lab: the overnight high, low and midpoint were touched and respected like random prices at the same distance from the open (721 sessions, Dec 2023 – Nov 2026, QQQ and SPY with futures-based overnight levels).

Prerequisite: M2.10 Multi-session structure · Next: M2.12 Shelf & ledge, which applies the same “describe, don’t forecast” reading to shelves inside a profile · Connects to: M2.07 (acceptance at a reference), M2.08 (overnight single prints), M2.09 (poor extremes), M4.06 (overnight inventory in session prep)

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