Delta profile

Friday 4 September 2026, 16:00 ET (03:00 ICT Saturday). The NQ regular session is over and you are writing Tuesday’s sheet (Monday is Labor Day). The busiest ten points of the day were 29,510 to 29,519.75: 32,560 contracts. More of them were sold into the bid than bought at the offer, and that row carried the most negative delta of the session. Price still closed at 29,568.25, about 48 points above it. Sellers leaned hardest at the busiest price and the day ended above them. Does that make 29,510 support for Tuesday? This lesson shows you how to read the profile, and what that kind of story is worth.

Key terms

Delta profile (hồ sơ delta): net delta (contracts bought at the offer minus contracts sold into the bid) at each price row over a period Definition. Bar delta (M3.04) tells you how much aggression happened; the delta profile tells you where.

Delta node (nút delta): a row with a large net reading, positive where buyers were the aggressors and negative where sellers were Definition. “Large” only means something against the session’s own baseline (below).

Term you need: volume profile. Volume at each price for the period; the busiest row is the POC and the busy clusters are high-volume nodes (M2.01–M2.04). The delta profile is always read next to it.

Chart first: one real session, two columns per row

The volume column says 29,510 was the day’s POC. The delta column adds who was more aggressive there: sellers, by 2,410 contracts. Price then finished above the row. That is the whole of the observation. Everything else people say about such a node is interpretation, and the interpretation is what this lesson grades.

How it works

Read the size of a node against the session

A delta node looks dramatic until you compare it with the rows around it. On 4 September:

The node was more one-sided than the day, but not by much: 46.3% of its contracts were buyer-initiated against 47.7% for the session. On a day when sellers were the net aggressors in 20 of 24 rows, a negative node at the busiest price is close to what you would expect by default. Before you call a node extreme, compute this ratio; ATAS can show delta divided by volume directly (the how-to below).

Two more limits come straight from the definition Definition:

• Delta is not position. −2,410 net does not mean 2,410 new shorts. Some of those sells were longs getting out, and delta says nothing about open interest. The node tells you where sellers crossed the spread, not how many of them are still holding.
• A node mixes hours. The 29,510 row traded in the 10:00 and 11:00 hours and again in the afternoon on 4 September. One number blends several separate visits to that price.

Four combinations, described

Read every node with its volume and the close. These describe the finished session Definition; none of them is a forecast.

• Busy row, delta near zero: both sides traded heavily and neither crossed the spread much more than the other. Two-sided trade.
• Busy row, one-sided delta, close on the other side: the aggressors there finished on the wrong side of the day’s close. The 4 September POC is this case.
• Busy row, one-sided delta, close on the same side: the aggressors there finished on the right side of the close.
• Thin row, large delta: a fast move through the price. Read it with M3.04’s same-interval rule: delta describes the move it belongs to.

Myth / hypothesis: “A busy row with heavy selling and a close above it shows absorption: the sellers are trapped, they will buy to get out, and the row becomes support. Delta extremes at the value-area edges are the strongest of these levels” Lore. The box below gives the closest tests. The trapped-trader story describes how a session could have unfolded; it has not been shown to tell you what the next session does.

What the data says

Why it is still worth knowing

• Shared vocabulary. “The POC was seller-heavy and we closed above it” is a precise, checkable description that other traders understand.
• Describing the auction. The profile shows where the session’s effort went, which a candle cannot.
• Framing risk. A busy, one-sided row is a price where many contracts changed hands; if you trade near it, expect activity and size your stop for it.
• A testable hypothesis. You can log every node like this one and score it yourself (the drill).

Worked example: write the line, then watch it

Sheet for Tuesday 8 September, written on Friday night:

“29,510–29,520 = 4 Sep POC, seller-heavy (−2,410, −7.4% vs −4.6% for the session), closed 48 points above. If price returns to the row and holds, with 5-minute closes back above 29,520, then I treat it as a place where buyers responded and look for a long with a stop below the row. If two 5-minute bars in a row close below 29,510, the read is wrong and I stand aside. Hypothesis, grade Lore.”

What happened, from the lab’s tape:

• Monday 7 September (Labor Day, shortened session): NQ never reached the row. The lowest trade from Sunday evening on was 29,530.50.
• Tuesday, 04:02 ET (15:02 ICT): overnight, price traded through the row to 29,477.75 at 04:12, then climbed back to the 29,600s by about 06:00.
• Tuesday regular session: open 29,644.50. At 09:42 ET (20:42 ICT) price was back under 29,520. The 5-minute bars starting 09:50 and 09:55 both closed below 29,510, so the plan said “stand aside” at 10:00. The low was 29,424.75 at 10:11 ET, 85 points under the row. The bar starting 10:20 closed back above 29,520, and the day closed at 29,542.75.

So the node neither stopped price nor marked a clean break. Your rule took you out at 10:00, and price then reclaimed the row within half an hour. One session proves nothing either way. It is simply what a no-edge result looks like on a chart: the level gets traded through and taken back, and the story fits whichever ending you look at afterwards.

The mirror and the failure

The short-side mirror is the same reading turned over: a busy, buyer-heavy row with the close below it. The popular claim is that those buyers are trapped and the row becomes resistance. The closest test, heavy buying at a high that fails, is inconclusive (box above), so the mirror is Lore too. Write it the same way: “if price returns and is rejected with closes back below, then …; if it accepts above, the read is wrong.”

Common mistake: Reading a node without the close: “big negative node, so bearish.” The node says who crossed the spread at that price; the close says where the day finished relative to it. Every node needs both, and even together they describe the session rather than predict the next one.

Common mistake (evidence): Treating “sellers lost at that price” as an edge. It is a description of yesterday. The closest tests show no edge for the support story, and the session delta-profile version has not been measured at all, so it stays a hypothesis you log, not a level you size up on.

Common mistake: Calling a node extreme without its baseline. On a seller-heavy day most rows are negative; compare the node’s delta-to-volume ratio with the session’s before you call it unusual.

Do it on the platform

Drill: score five delta nodes

For five sessions, take the most one-sided busy row of the regular session (among the top five rows by volume, the one with the largest absolute delta).

1. Log the date, row, volume, delta, delta as % of volume, the session’s delta as % of volume, and where the close sat relative to the row.
2. Before the next session, write the if-then line from the worked example, with the price that proves it wrong.
3. Next session, score the first return to the row: held (closes back on the close’s side), broke (two closes beyond it), or not reached.

What we don’t know yet

• Whether session delta-profile nodes, scored the way the drill does, hold more often than busy rows with balanced delta: not yet measured on NQ in our lab. The closest public tests are the bar-level and prior-POC tests above.
• Whether the delta-to-volume ratio, rather than raw delta, separates nodes that matter: not yet measured.

Recap:

• The delta profile is net aggression by price: it shows where buyers or sellers crossed the spread.
• Read every node with its volume, the close and the session’s own delta-to-volume ratio.
• A seller-heavy busy row under an up-close describes the session; as future support it is Lore (the closest NQ tests show no edge).
• Write a node as an if-then line with its invalidation, and score it, rather than trading it on sight.

Further reading:

• Order Flow, Measured: the public NQ practitioner study behind the evidence box (claims FP-12, VP-05, DA-06).
• Cumulative Volume Delta / CVD Divergence: delta and CVD used as confirmation at known levels, not as standalone entries.
• The Ultimate Guide To Profiting From Footprint Charts: footprint delta with worked ES examples; read its trading claims against the evidence box.
• Our lab: the 4 and 8 September 2026 NQ sessions above, measured from our tick recording.

This lesson connects to: M3.04 (delta: the input, and why it describes the same interval rather than forecasting) · M3.02 (absorption: the story this profile is used to tell) · M2.01 (volume profile: the required companion read). Next, M3.07 moves from where volume traded to its volume-weighted average price.

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