Poor highs & lows

It is 12:30 ET (00:30 ICT, winter) and NQ’s high of the day sits at 18,060. Two different 30-minute periods traded up there and stopped at the same price: a flat top, no tail. Profile traders call that a poor high and say price “owes” it a revisit. Yesterday’s high, by contrast, had a long spike that was sold hard, and they call that one “defended”. Both stories sound like edges. This lesson gives you the definitions, a way to read a developing profile with them, and what happened when the two stories were tested.

Every extreme gets one question, the completion test: did the auction finish its business here? The answer comes from counting rows at the extreme Definition:

• A tail (đuôi) is the run of prices at a profile extreme that each hold exactly one TPO (one 30-minute letter), counted inward from the extreme. A selling tail sits at the high, a buying tail at the low.
• Excess (excess, đuôi từ chối) is a tail of at least two rows. In auction language the market traded too far, found no business there and left quickly: the auction at that extreme looks complete.
• A poor high/low (đỉnh/đáy kém) is an extreme without excess: a flat top or bottom where two or more TPOs share the extreme price, or a tail only one row long. In auction language the auction stopped there without showing that it had finished.

These are descriptions of what already traded. Whether either one tells you where price goes next is a separate question, answered below.

Chart first: a poor high that turns into excess

After period E, the top row has two letters. Nobody traded above 18,060, and nobody was rejected from there quickly either; two periods simply ran out of buyers at the same price. That is what “poor” means: incomplete, not weak. After period H, the two rows above 18,060 each hold one letter. Price got there, found no one willing to trade more, and left inside one period. That is excess.

All of this is visible on the developing profile. You do not need the close to see a flat top forming at 11:30 ET, and you do not need it to see a tail completing it later. One caution: a developing extreme can change every period, so the label is “so far” until the session ends.

How it works

The auction story, and where it stops being fact

Auction market theory reads a tail as a completed probe: the market searched beyond fair value, found no trade, and came back. A flat top reads as a search that stopped without that answer. That much is description, and it is useful.

The next step is where the story becomes a claim about the future.

Myth / hypothesis: “A poor high is unfinished business, so price has to come back and repair it; an excess tail traps the traders who chased the spike, and their forced exits defend the level on the next test.” Profile traders have taught both halves for decades. When tested, poor extremes were revisited no more often than tailed extremes or random prices at the same distance, and tailed extremes held no better than random prices Lore. The trapped-trader mechanism has not been measured at all.

Why it is still worth knowing

The completion test gives you shared vocabulary with every profile trader and every later lesson in this course. It describes the auction honestly: “the session stopped at 18,060 without a tail” is a fact you can journal. It frames risk: if your read is that a top is finished, the far side of the tail is where that read is wrong. And “poor extremes get revisited” is a clean hypothesis you can score on your own charts against random prices, which is exactly what the tests below did.

What the data says

In short: our lab’s test is a measured null on the index proxies (QQQ, SPY), not on NQ futures Measured: no edgeNasdaq index, and the NQ practitioner result points the same way. Neither label has shown that it moves the odds of a revisit or a hold.

Worked example: reading the session in the figure

Synthetic NQ session, 30-minute letters, 4-point rows.

09:30–10:30 ET (21:30–22:30 ICT). Periods A and B drive up from 18,016 to 18,044. A’s first two rows hold only A: a two-row buying tail, so the low is excess. B covers 18,032–18,040 alone: single prints (M2.08), a record of where the morning moved fast.

10:30–12:00 ET (22:30–00:00 ICT). C, D and E push to 18,060 and stop. By 11:30 ET the top price holds D and E: a flat, poor high. F and G then rotate between 18,044 and 18,056.

The read at 12:30 ET. “The auction at 18,060 stopped without a tail. That describes the morning; it does not forecast the afternoon.” What you write down is conditional: if price comes back to 18,060 and builds time and volume above it, buyers are being accepted higher; if it spikes through and leaves fast, the top is being completed with excess. Either answer is information. Neither is owed: the evidence above found poor extremes revisited no more often than random prices.

13:00 ET (01:00 ICT). Period H trades up through 18,060 to 18,068 and falls back to close the period at 18,048. Rows 18,064 and 18,068 hold only H: a two-row selling tail. The high is now excess, and the profile says the auction at the top looks complete.

What that buys you. A clean journal line and a reference for risk. If you are short later in the session for your own reasons, “back above 18,068 with time building” is the price where your read of a finished top is wrong, so your stop belongs beyond it rather than inside the tail. What it does not buy you is a level that holds: tailed extremes held no better than random prices in both tests. The next test of 18,068 is a question for the auction, not a fade with built-in fuel.

The mirror and the failure

Short side. A poor low is the same description at the bottom: two or more TPOs at the session low, or a one-row tail. A buying tail of two rows or more makes the low excess. The evidence boxes cover both sides; highs and lows were pooled.

Where the story fails. Take the same session. Next morning NQ opens at 18,130, above the whole profile, and trends for three days without trading back. The excess high at 18,068 was never tested, and nothing about it mattered. Or the next day opens at 18,040, rallies through 18,068 in period B and spends two hours above it: the “completed” top is accepted through, and anyone fading it because of the tail was run over. And many poor highs sit behind price for weeks; in our lab’s test a random price at the same distance was touched about as often.

Common mistakes

Common mistake: Treating a poor high as a target: “the auction has to come back, so I hold my long for 18,060.” Nothing has to come back. A target chosen because of a claim that tested no better than random prices adds hope to a trade, not edge. If you want a target, pick it for a reason you can score, and score it.

Common mistake: Treating an excess tail as a wall: fading every retest of a tail because “trapped traders defend it”. The trapped-trader story is a hypothesis nobody has measured, and tailed extremes held no better than random prices. A tail tells you where your read is wrong; it does not tell you that the read is right.

Common mistake: Reading “most poor highs get revisited” as an edge. Most prior extremes sit close to the next session’s price, and nearby prices get touched often whatever their shape. A revisit rate means nothing until you compare it with random prices at the same distance (EV1.02).

Drill: score the revisit on your own charts

Take 20 prior RTH sessions of NQ or ES on 30-minute TPO.

1. For each session’s high and low, count the single-print rows at the extreme: two or more is excess, anything less is poor. Write the label before you look at the next day.
2. Record whether the next session traded at that extreme price (yes/no).
3. For each extreme, pick a control price at the same distance from the next session’s open, on the other side, and record whether that price was touched too.
4. Compare: revisit rate of poor extremes, of excess extremes, and of the controls.

What we don’t know yet

• NQ futures with true 30-minute letters over many years: not yet measured by our lab (a long NQ intraday history is still the data gap); our test used hourly letters on the ETF proxies.
• Overnight (ETH) profiles: not yet measured.
• Small effects under about 3–5 pp, and discretionary use that combines the label with order flow at the extreme, are not ruled out by either test.

Recap

• The completion test counts single-print rows at the extreme: two or more is excess (the auction there looks complete); a flat extreme or a one-row tail is poor (it stopped without finishing).
• Both labels are readable on the developing profile, and both can change until the session closes.
• “Poor extremes get revisited” and “excess tails hold” were tested on NQ and on the index proxies and showed no edge; treat both as hypotheses, not as targets or walls.
• Use the labels to describe the auction and to place the price where your read is wrong, then score any revisit or hold idea against random prices at the same distance.

Further reading

• Mind Over Markets — James Dalton: the auction-market text behind tails, excess and poor extremes, as definitions.
• Following the Auction: Market Profile Basic Language: plain-English vocabulary for reading extremes.
• Order Flow, Measured: one practitioner’s public, pre-registered NQ study; claims AP-05 and AP-06 cover poor extremes and tails.
• Our lab: poor extremes revisited and excess extremes held no more often than random prices at the same distance on QQQ and SPY (TPO concepts test, Dec 2023 – Nov 2026).

Prerequisite: M2.08 Single prints · Next: M2.10 Multi-session structure, where extremes and balance areas are read across several sessions · Connects to: M2.08 (singles inside the profile), M2.11 (poor overnight extremes), M1.07 (trapped traders, kept as a hypothesis here)

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