Heatmap & liquidity walls

It is 10:05 ET (22:05 ICT, winter). ES has ground up from 5,990.00 to 5,999.00 in half an hour, and on the heatmap a bright orange band sits at 6,000.00: about 1,000 contracts offered, more than five times anything near it. Every lift of the offer at 6,000.00 trades a few hundred lots and the band stays bright. You think “resistance” and lean short at 5,999.00.

The DOM (M3.08) showed you a snapshot of that offer. The heatmap shows you its history. This lesson is about what that history tells you for certain, what it plausibly tells you, and what is popular but untested.

Key terms

Heatmap (bản đồ nhiệt): a chart of the order book over time. Each column is a moment, each row a price, and the brightness of a cell is the displayed resting size at that price at that moment Definition. It is the DOM recorded as a movie.

Liquidity wall (tường thanh khoản): a bright horizontal band, meaning unusually large displayed size resting at one price for a stretch of time Definition. “Unusually large” is a setting you choose, for example several times the median size of the nearby levels.

Void (vùng trống): a dark stretch of the heatmap where little size is displayed Definition.

Term you need: displayed size vs traded size. Displayed size is what the book advertises; it can be cancelled at any moment at no cost. Traded size is what actually changed hands, from the tape (M3.09). The heatmap shows the first; bubbles or the tape show the second. Most of this lesson is about not confusing them.

Chart first: one wall, three phases

The same story as a sketch, with the times used in the worked example below (synthetic):

6,004 ── ·······························  void: little size displayed
6,002 ── ·······························
6,000 ════════════════╗                   wall: 800–1,200 contracts shown
5,998 ── ────────────╱│ 10:05–10:45 ET    traded into and refilled
5,996 ── ·············│ 10:45–10:46:30    faded, almost nothing traded
                      │                   price 6,004.00 by 10:47

How it works

The three fates of a wall

What happens to a wall when price comes to it is the part you can read. Its size alone tells you little, because size costs nothing to show.

1. Traded and refilled. Contracts print at the wall’s price and the displayed size comes back. This shows for certain that real size traded there, as in absorption (M3.02) Definition. It does not show that the wall keeps refilling: “iceberg orders at a level make it hold” was tested on NQ and came out inconclusive (box below) Lore.
2. Withdrawn before contact. The size disappears as price approaches, before much trades. It provided no defence. The intent is unknown: most pulled orders are plausibly routine quote management by market makers, not spoofing (M3.12) PlausibleOne trader's study.
3. Faded without trading. After a period of holding, the size drains away with little volume printing there. The defence is gone. Whether a draining wall predicts a break has not been tested by anyone we can cite Lore. In the sketch price did rise; on another day the same fade is followed by nothing.

The honest reading of all three is about the present: is the size that your idea relies on still there, and has any of it actually traded?

Persistence within the session

One claim about walls has a mechanism and a reasonable chance of being true: a band that has stayed on the book for hours is more likely to still be there when price arrives than one that flashed up a minute ago PlausibleOne trader's study. Short-lived quotes are mostly algorithms adjusting; size that has sat through several swings is more likely to belong to someone who wants to trade at that price. Notice what the claim does not say: nothing about whether the wall stops price when it gets there. It has not been measured on NQ or ES; the box below explains why.

Voids

With little size displayed, each aggressive contract can move the price further, so moves tend to travel faster through dark areas PlausibleOne trader's study. This is the depth version of a low-volume node (M2.03); on the volume profile, the “price moves faster through LVNs” claim leans the right way in one trader’s NQ study but is inconclusive. On depth it is not yet measured. Displayed size can also appear in a void at any moment, so read a void as “less friction if nothing changes”, not as a forecast of speed.

Myth / hypothesis: “Liquidity has memory: yesterday’s walls reappear at the same prices tomorrow, because the same participants post the same size every day.” Lore Nobody has tested next-day reappearance of walls on NQ or ES that we can cite, and the lab has no recorded depth history to test it on. Checking yesterday’s heatmap at your levels is a fine habit, as long as you score it: note the price, whether a wall appeared there the next day, and what happened at contact.

Myth / hypothesis: “Big walls usually get pulled before price arrives, and a wall that drains means a break is coming.” Lore Both are popular, and both were listed as untested in the public NQ study because it had no depth data. A rule built on either is a hypothesis you could test, not a fact to trade.

Why it is still worth knowing

The heatmap is a good description tool. It tells you where size is advertised now and where it was, it shows you when the size your trade depends on has left, and it gives you shared vocabulary (wall, void, pulled, refilled) to describe the auction in your journal. Those uses need no forecast. What it does not give you, on the evidence available, is a tested rule of the form “wall, therefore price turns”.

What the data says

Worked example: right for 40 minutes, run over in 90 seconds

ES, synthetic morning, as in the sketch. You short 1 contract at 5,999.00 at 10:05 ET because the 6,000.00 offer has traded and refilled for an hour. Your written plan: stop 6,001.50 (10 ticks, $125), target 5,993.00, and exit if the offer at 6,000.00 is gone House rule.

For 40 minutes the plan works the way you read it: lifts trade into the wall, it refills, price cannot get through, and you are a few ticks green. At 10:45 the band starts to fade. By 10:46:30 it shows under 200 lots, and the tape shows only about 60 contracts traded at 6,000.00 in that time. The size was withdrawn, not overrun.

Your premise was “real supply at 6,000.00”. That premise is now off the screen, so the plan says exit. You buy back at 5,999.75, a small loss of 3 ticks ($37.50). Price trades 6,004.00 by 10:47, which would have stopped you at 6,001.50 for $125 plus slippage.

The point is not that the fade predicted the move (that is untested). The point is that your reason for being short left, and the exit rule did not need a forecast to act on it. The trader who kept the short “because 6,000 is resistance” was trading a photograph of something that no longer existed.

The mirror and the failure

Mirror (long side). A bid wall of about 900 contracts sits at 5,980.00 and has traded and refilled through two dips. You buy 5,981.00 with a stop at 5,978.50 and the same rule: exit if the bid at 5,980.00 is gone. The logic is identical; only the side changes.

Failure. Same short, another day. The 6,000.00 wall fades at 10:45 with little trading, you exit at 5,999.75, and price drifts down to 5,994.00 over the next twenty minutes. The fade was not a break signal at all. Your exit cost you the trade’s profit, and that is fine: the rule protects you from holding a position whose evidence has gone, not from every missed winner. Track both outcomes in your journal before you decide the rule is worth keeping.

Do it on the platform

OrderflowSuite has no depth heatmap of its own. Its “Delta Heatmap” study colours traded delta, not resting size; do not confuse the two (PLAT3.08).

Prop-firm note. Evaluation accounts fill your orders in a simulator, which cannot know your real place in the queue in front of a wall. Firms treat strategies that exploit simulated fills as prohibited conduct; the Prop Firm course (PROP1.04) will quote the rule with its date once we have checked it.

Common mistakes

Common mistake: Treating a bright wall as a permanent level: “1,000 offered at 6,000, so short.” Displayed size is the cheapest thing on your screen to change. The size is not the level; the trading and refilling are the only part that shows real interest, and even that is not a tested forecast.

Common mistake: Reading a plausible claim as a measured edge. “Persistent walls are more likely to still be there” is graded Plausible and says nothing about whether price turns there. Restating it as a rule that persistent walls stop price turns a description into an untested forecast.

Common mistake: Reading intent into a pulled order. A wall that steps away was, in most cases, routine quote management. Calling every pull a spoof (M3.12) builds a story the screen cannot confirm.

Drill: score wall fates for five sessions

Use ES or NQ, RTH only, on replay or live.

1. Fix your wall definition (for example, at least 3 times the median nearby size, at least 3 ticks from the inside) and write it down.
2. Each time price comes within 2 ticks of a wall, log: time (ET), price, how long the wall had been on the book, its fate (traded and refilled, withdrawn, faded), and what price did over the next 5 minutes (through, back, nowhere).
3. After five sessions, compare the walls older than one hour with the younger ones: what share of each was still there at contact?

What we don’t know yet

• How often walls on NQ or ES are pulled before contact, by wall age: not yet measured (the lab’s depth tests wait for a recorder).
• Whether draining depth or deep-book imbalance forecasts anything beyond the top-of-book queue imbalance taught in M3.08: not yet measured.
• Whether walls reappear at the same prices the next day: not yet measured.

Recap

• A heatmap is the order book recorded over time; brightness is displayed size, which can vanish at no cost.
• Read a wall’s fate: traded and refilled shows real size traded; withdrawn or faded means the defence is gone, with intent unknown.
• Within a session, a long-lived wall is plausibly more likely to still be there at contact (Plausible, one trader’s study); that is not a claim that price turns there.
• Next-day reappearance, “big walls get pulled” and “a draining wall means a break” are untested lore.
• Write the wall into your plan as a condition and exit when it is gone.

Check yourself

Further reading

• Order Flow, Measured: All claims. A public, pre-registered practitioner study on NQ, 2018–2026. Read the order-book rows (OB-01 to OB-07) and the heatmap note (DN-19); most depth claims are marked untested because the study had top-of-book data only.
• Cont, Kukanov and Stoikov (2014), “The Price Impact of Order Book Events”, Journal of Financial Econometrics 12(1). Order-book events explain price changes over the same interval; explanation, not forecast (EV3.03).
• Gould and Bonart (2016), “Queue Imbalance as a One-Tick-Ahead Price Predictor in a Limit Order Book”, Market Microstructure and Liquidity 2(2). Measured on stocks, and only at the top of the book.
• Our lab: depth and wall claims are not yet measured; a depth recorder and about twenty recorded sessions are queued before any test.

Next: M3.11 moves from displayed size to traded size: bubbles and sweeps, and what a large print does and does not tell you.

This lesson connects to: M3.08 (DOM, the snapshot this turns into a movie) · M3.12 (spoofing vs routine cancellation, the two readings of a pulled wall) · M4.07 (level map, where your heatmap notes get scored).

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