Composite profiles

It is 09:00 ET (21:00 ICT, winter), before the open. Yesterday’s NQ POC was 18,012. A profile of the last twenty sessions puts its POC at 17,940, with a thick band of volume from 17,920 to 17,960. Which belongs on your chart, and is the month’s level “stronger”? This lesson builds the bigger picture and shows what the evidence says about trusting it.

Key term

A composite profile (hồ sơ tổng hợp) merges the volume at each price from several sessions into one distribution: 5 days, 20 days, or a custom period Definition. It has the same parts as a single-session profile (M2.01–M2.03): a POC, a 70% value area, high-volume nodes (HVNs) and low-volume nodes (LVNs). The difference is the question it answers: “where has the market done business across days?” instead of “where did it trade yesterday?”

A composite is a record, not a crowd: it shows that many contracts traded at a price, not who traded them or whether they are still in.

Chart first: twenty sessions in one profile

The heavy band from 5,940 to 5,970 is where the month’s business happened: the composite value area, built as in M2.02. The thin band from 5,975 to 5,990 is a composite LVN: prices the market passed through without doing much business.

Today’s single-session levels now have an address. “Price is below the 20-day composite POC but holding today’s VAL” is a complete location sentence, the quantified version of the M1.06 trade sheet.

Use composites for context and single-session profiles for timing: the composite tells you which prices to watch; today’s profile and the order flow at the level tell you what is happening there now.

Choosing the lookback

Choosing the lookback is choosing your question. Intraday traders ask “where has value been this week?”: 5 to 10 days. Swing traders ask “where has value been this quarter?”: 20 to 60 days. These ranges are conventions, not tested settings House rule. Too long, and this week’s migration is buried under old history; too short, and one session’s noise looks like structure.

There is a better option than a round number: anchor the composite to a balance. If the market rotated between 5,935 and 5,975 for eleven sessions, build the composite over exactly those eleven sessions. Its POC, value area and edges then describe one auction, not a calendar slice that mixes a trend leg with a balance. When price leaves the balance, start a new composite and keep the old one as the map of where value was (M2.10).

Practical traps: rolls, sessions and “decay”

Contract rolls. ES and NQ roll quarterly (March, June, September, December). The next contract trades at a different price from the expiring one because of carry: financing cost minus dividends (M0.08). In September 2026 roll week our lab saw the next NQ contract about 230 points above the expiring one MeasuredNQ. A 20-day composite that stitches the two contracts without adjusting puts part of its volume at prices that are a few hundred points off. Use a continuous contract your platform adjusts correctly, or start the composite after the roll.

Session definition. An ETH composite includes overnight volume; an RTH composite does not. Their POCs can differ. Pick one and keep it, as with single-session profiles (M2.01).

Decay is your job. Most composites are plain sums: a contract traded 40 sessions ago counts exactly as much as one traded yesterday Definition. The software does not fade old volume; you do, by choosing the lookback. If an old HVN dominates but price left it weeks ago, shorten the window or anchor a new composite.

The popular claim, and what was tested

The usual teaching says composite levels outweigh single-session levels because “weight follows volume × time”: more business over more sessions, more traders with a reference price there. Tested, it showed no edge.

Myth / hypothesis: “Multi-week composite nodes react more than single-session nodes” Lore. On NQ a public practitioner study found −1.6 percentage points (95% CI −12.4 to +9.2), n = 1,047: no advantage shown, and the interval is too wide to rule out a sizeable effect either way. Treat composite levels as places to watch, not as stronger levels.

Myth / hypothesis: “HVNs act as magnets” Lore. The same study found +0.8 points (CI −0.1 to +1.7), n = 10,908: an edge of 3 points is ruled out. Nothing suggests a composite HVN pulls harder.

The one node idea with a weak lean in its favour is LVN speed: on NQ, price moved faster through LVNs than through HVNs in the same study (+7.9 on its own scale, CI +0.5 to +15.8, n = 794), but it did not survive the study’s correction for multiple tests PlausibleOne trader's study. Whether composite LVNs behave the same way is not yet measured.

Why it is still worth knowing. A composite gives you words for the larger auction: where the month’s business was, where it was thin, and where today sits. It frames risk and stops, and it turns “the market feels high” into prices you can write down and score. Each expected behaviour here is a hypothesis you can test in your journal.

What the data says

Worked example: discovery above the composite

Synthetic ES example. The 20-day composite above: POC 5,950, value area 5,940–5,970, composite LVN 5,975–5,990. Today trades 5,995–6,010, above the entire composite.

Read: the market is trading at prices it has not negotiated this month. There is no composite volume overhead; below sits first the thin zone at 5,975–5,990 and then the value area at 5,940–5,970. You buy 1 ES at 6,005 on a pullback that holds today’s developing value. The composite helps you write the plan:

• Stop. 5,992, just above the thin zone: 13 points, $650 on 1 ES ($50 a point) before costs. Back inside the thin zone, the reason for the trade (“holding above the month’s range”) is gone.
• If it fails. Expect little trade in the thin zone and two-way business in the value area. Both are hypotheses: LVN speed is a weak lean on NQ PlausibleOne trader's study, and no study shows the value area stops the move.
• What it does not say. Whether 6,005 holds. That comes from today’s profile and the order flow at the level (M3, M4.09).

The composite turned a vague “overbought” into prices with questions: what happens at 5,990, and does value start building back inside 5,940–5,970?

Worked example 2: the composite that mixed two auctions

Synthetic NQ example. A trader uses a rolling 20-day composite. The first eight of those days were a trend leg from 17,600 to 17,900; the last twelve were a balance between 17,880 and 17,980. The composite shows a long, thin lower half and a fat upper half, with its POC at 17,930.

Read: the window blended two auctions. The thin lower half is the trend leg’s path, not a gap between two balances. Anchored to the twelve balance days, the picture is cleaner: value area 17,900–17,960, POC 17,932, edges at 17,880 and 17,980, the prices a breakout has to clear. Before you trust a composite’s nodes, ask whether the window covers one auction or several.

Common mistake: Treating the composite POC as a target: “price has to return to the 20-day POC.” In the same public NQ study, untouched POCs were revisited 85% of the time and random prices at the same distance 84% of the time: no edge shown. In a migrating market price can spend weeks above a composite POC while the composite drags upward behind it; fading every leg back to it is how traders short a trend.

Common mistake: Reading the evidence box as “composites are useless.” No edge shown is not proven useless; the composite is still the cleanest map of where the larger auction did business.

What we don’t know yet

• Whether composite LVNs are crossed faster than single-session LVNs on NQ: not yet measured.
• Whether a composite POC or value-area edge is respected more often than a random price at the same distance: not yet measured.

Recap

• A composite merges several sessions’ volume into one profile; read its POC, value area, HVNs and LVNs the same way as a single session’s.
• Composite nodes have not been shown to react more than single-session nodes (no edge shown, one trader’s study); use them as context and as prices to watch.
• Match the lookback to your horizon (intraday 5–10 days, swing 20–60), or better, anchor it to one balance so it describes one auction.
• Composites do not decay on their own, and contract rolls and session definition can shift every level: check the data before trusting the map.

Further reading

• James F. Dalton, Eric T. Jones and Robert B. Dalton, Mind Over Markets (2nd ed., Wiley, 2013), chapters on value and multi-day auctions. Read the definitions, then ask how each claim would be counted.
• Order Flow, Measured: All claims. A public practitioner study on NQ; rows VP-07 (composite vs single-session nodes), VP-02 (LVN speed), VP-03 (HVN pull) and VP-01 (untouched POCs).

Prerequisites: M2.02 Value Area High & Low; M2.03 High & Low Volume Nodes · Next: M2.05 Profile shapes, which reads the shape of a profile, single or composite, as a description of how value formed · Connects to: M0.08 (basis and the roll), M2.01 (single-session POC), M2.10 (multi-day balance and breakouts), M1.06 (the location sentence), M4.07 (building a level map)

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