Skip to content

What order flow is and what it records

By Author of ORDER FLOW — Market Mechanics

A price chart shows where the market traded. Order flow shows how it got there.

That line gets repeated a lot, usually with a promise attached that it cannot keep: that reading the flow lets you see the next move coming. Order flow is a record of transactions that have already happened, and a record does not contain the future.

What it does contain is information the candlestick chart throws away when it compresses a whole auction into four prices: who executed, against whom, at which level and in what size.

Who moves price

At every price level on NQ there are two kinds of participant. Some wait. They have left a limit order in the book and are in no hurry, and if price never comes to them the order never fills. Others do not wait: they trade now, at whatever price is on offer.

To understand what moves price, the distinction that matters is between waiting and executing, whichever side of the market each participant is on. The four combinations, and why a footprint is built from only one of the two groups, are set out on the page about what a footprint chart cannot show.

Waiting orders make up the available liquidity and executing orders consume it. Price only moves once the resting liquidity at a level is gone, and that can happen in two ways. Aggressive orders can use it up, contract by contract, and then every execution is recorded. Or the passive side can pull it before anyone reaches it, and then price advances without a single trade.

A · Someone pushes Bid Ask 21,302.00 24 88 21,301.75 31 142 21,301.50 27 165 21,301.25 35 151 21,301.00 40 63 Volume 766 Delta +452 Aggressive buying at every level. The advance left a trail. B · Nobody resists Bid Ask 21,302.00 6 9 21,301.75 4 7 21,301.50 5 3 21,301.25 7 8 21,301.00 9 6 Volume 64 Delta +2 Hardly anything traded on either side. The offers left, or were never there. Open, high, low and close are identical. The record at each level is not.
Schematic of one NQ candle recorded two ways: bought level by level, or left unresisted. Values are illustrative.

On a candlestick chart the two cases look identical. In the schematic, the first candle trades 766 contracts with a delta of +452. The second has the same open, high, low and close and trades 64. One advance leaves a trail in the record, and the other leaves almost nothing.

So what is order flow?

Order flow is the record of aggressive execution against passive liquidity: who crossed the spread, in which direction, in what size, at which price and when.

An indicator is calculated from price and tuned with parameters. Order flow is collected. Every number on the screen stands for a transaction that already took place between an aggressor who executed and a passive order that was waiting at that price.

That is what makes possible a reading the price chart cannot give: telling a move that happened because someone pushed from a move that happened because nobody resisted. In the first there is the conviction of someone committing capital. The second is a simple absence of friction, and one order on the other side is enough to reverse it.

Four views of the same flow

Much of the early confusion comes from treating the DOM, the tape, the footprint and the volume profile as four techniques to be learned separately. They are four representations of one flow, laid out on different axes.

DOMTapeFootprintVolume profile Resting liquidity. What has not executed yet. Executions one by one, in time order. Executions per level, inside each candle. Total volume per level, across the period. Four representations of the same flow, each on a different axis.
Schematic of the four usual representations of the flow, side by side. Proportions are illustrative.

The DOM, or depth of market, is the only one of the four that shows orders that have not executed yet: pending intent, the liquidity waiting on both sides of the spread. It is also the most fragile, because that liquidity can be withdrawn in milliseconds and is sometimes there precisely to be seen.

The tape records executions one at a time, in the order they happened. It is the rawest data and the hardest to read live, because too much happens every second on NQ.

The footprint groups those same executions by price level inside each candle, split by whether they hit the bid or lifted the offer. Most of the analytical work happens here, and it is the view that asks for the most care: comparing the two columns of one level gives that level's delta, and who won the cross is read on the diagonal.

The volume profile gives up the detail of who was aggressive and keeps the total volume traded at each level over a period. It loses resolution and gains context, showing where the market has accepted price and where it has rejected it.

Specific concepts are then read on top of these views: delta, the net difference between buying and selling aggression; absorption, when heavy aggressive volume barely moves price; imbalances between adjacent levels; and exhaustion, when aggression fades at the end of a move. Each one has its own mechanics and its own false positives.

Why it does not predict

Take an NQ level where 1,240 contracts of aggressive buying execute and price cannot get above it. The record leaves no doubt: someone with size was buying there, and someone with size was selling passively against that buying. That much has already happened.

The record Bid Ask 21,302.25 40 55 21,302.00 210 1,240 21,301.75 180 300 21,301.50 220 190 1,240 contracts bought at the ask at 21,302.00. Price does not advance. Outcome A · continues Buying aggression finds relief and the level holds. Outcome B · reverses The buyer stops showing up and the level gives way. The record is identical in both cases. The outcome is not contained in it.
Schematic of one absorption record followed by two possible price paths. Values are illustrative.

What the record does not hold is what either of them does next. If the passive seller keeps selling, price gives way. If the seller runs out of inventory or steps aside, price carries on. Both paths fit exactly the same record, because it describes what was done and says nothing about what anyone intends to do.

Readings also go stale. An absorption five minutes ago does not mean the participant behind it will still be there five minutes from now. Positions get closed and algorithms change regime. The market at 10:35 ET is a different market from the one at 10:30, even if the chart has only added two candles.

So “there is evidence that a participant with size is active on the buy side at this level” is a more accurate sentence than “price is going up”. Both describe the same situation, and they lead to opposite behaviour. The first invites you to keep watching and lets you get out as soon as the evidence is invalidated. The second invites conviction, and costs more when it is wrong.

What the record leaves out

Four limits are structural, and no platform or data vendor removes them. The page on the limits of the footprint goes through each in detail. In short:

Liquidity that never traded is invisible. The flow records crossings. Limit orders that waited and were never touched leave nothing behind. The DOM shows them while they are there and keeps no memory of them afterwards.

The participant's real size does not appear. An iceberg order shows a fraction of itself and refills as it fills, so the record shows many small executions where there was a single large decision.

Intent is not recorded. Knowing that 1,240 contracts were bought does not tell you whether someone was opening a position, getting out of a short or offsetting risk held elsewhere. Every one of those produces the same number.

A data point on its own means nothing. The same imbalance can be the cleanest signal of the day near the border of value after a quiet pullback, and routine noise inside a zone where price has rotated all morning. Nothing in the cell says which.

That last limit destroys the most capital, because the flow keeps producing signals that are technically real and operationally irrelevant. The tool measures activity. Whether that activity means anything depends on a context the tool does not contain.

Why futures, and why NQ

Flow analysis needs a condition that not every market meets: the volume has to be real and centralised.

On CME futures every trade in a contract goes through the same book. The volume you see is the volume that traded, and every execution has a price and a side that can be checked. Spot forex has no single book, and the volume a CFD platform shows belongs to that provider's clients, not to the market. A good share of what circulates about order flow applies the analysis to those settings, where the conditions it depends on do not hold.

Within futures, NQ has a character of its own. Its tick is 0.25 index points, worth $5 according to CME's contract specifications, execution is fast, and price tends to travel several levels at once when liquidity runs out. Its imbalances and absorptions cannot be read with the same thresholds as those of a deeper contract such as ES, for reasons set out in why NQ imbalances do not read like ES imbalances.

The three most common misreadings

Taking delta for direction. Delta measures the imbalance between aggressors, and price gives the direction. A strong positive delta with price stuck is buying being absorbed, which often points the other way.

Treating high volume as a signal. High volume only says there was activity. A high-volume zone often sees a lot of trading precisely because both sides are content to do business there, with neither one getting the upper hand.

Reading the flow without a map. This is the error behind all the others. Look at a footprint without first deciding which levels matter and you will find signals everywhere, because a normal NQ session produces dozens of them. The flow helps you decide what to do when price reaches a place you had already decided to watch.