Order Flow, Explained Properly

A candlestick tells you four numbers: open, high, low, close. It cannot tell you the thing that produced them — whether that green candle was built by aggressive buyers lifting offers, or by sellers quietly absorbing every one of them until they ran out of buyers to sell to. Those are opposite situations and they look identical on a candle.

Order flow is the attempt to see inside the bar. This guide covers what the tools actually measure, how to read a footprint chart, what delta and imbalance are really telling you, and the substantial list of things order flow is routinely claimed to do and does not.

The one distinction everything rests on

Every trade has a buyer and a seller. “More buyers than sellers” is not a thing that can happen. What can differ is who was in a hurry.

• A limit order rests in the book and waits. It is patient. It provides liquidity.
• A market order crosses the spread to trade right now. It is aggressive. It takes liquidity.

Order flow analysis classifies every executed contract by which side crossed the spread. A contract traded at the ask was bought aggressively; one traded at the bid was sold aggressively. That single classification is the raw material for everything below — delta, imbalance, absorption, all of it.

It also immediately tells you the first limitation: the passive side is invisible. You see who was impatient, never who was waiting.

The footprint chart

A footprint (also called a cluster chart or bid×ask chart) splits a single price bar into rows, one per price, and shows the aggressive volume on each side at each price.

Two different comparisons are happening in that picture and it is worth separating them now, because mixing them up is the single most common mistake in reading a footprint.

Delta is horizontal. At each price, ask volume minus bid volume. It answers “at this price, which side was more aggressive?”

Imbalance is diagonal. The ask at one price is compared against the bid one tick below, and the bid at one price against the ask one tick above. That is explained in the next section, and it is not a stylistic choice — it is forced by how the order book works.

Reading the rows in sequence is the skill. In the diagram above, the heavy selling sits at the bottom of the bar and the heavy buying at the top, with three stacked buy imbalances carrying price into the highs — a bar in which sellers were aggressive into the lows and buyers then took control. The same bar drawn as a candle would be a single green candle, and would tell you none of that.

Delta

Delta is aggressive buying minus aggressive selling, over a bar or a session. Positive means market buyers were more active; negative means market sellers were.

Cumulative delta runs the total across the session, giving a curve you can compare to price. That comparison is where the useful and the overclaimed both live.

The genuinely informative case is divergence: price makes a new high while cumulative delta does not. Read literally, that says the new high was achieved with less aggressive buying than the previous one — the buyers pushing are running out, or the sellers meeting them are getting bigger. That is real information.

What it is not is a signal. Delta divergence appears constantly in trending markets and resolves in the direction of the trend most of the time. On its own it is a reason to look, not a reason to act, and traders who treat it as an entry trigger spend a lot of money learning that distinction. It is also mechanically noisy in thin conditions, where a single moderate order swings the number.

Imbalance

Here is the part that trips people up. Imbalance is not the two numbers sitting side by side in one row. It is diagonal: the volume bought at the ask at one price, against the volume sold into the bid one tick below.

The reason is mechanical. At any instant the best ask and the best bid are one tick apart — that is what the spread is. So a buyer lifting the offer at 5041 and a seller hitting the bid at 5040 are the two flows that were genuinely competing with each other at that moment. Comparing the bid and the ask printed at the same price compares two things that happened at different times against different counterparties, which is why platforms do not do it.

When one side of a diagonal pair exceeds the other by a chosen ratio, that cell is flagged as an imbalance. Thresholds vary: 200% and 400% are the documented defaults in the major platforms, 300% is a common middle setting, and some tools let you switch the comparison to horizontal — which is worth knowing about mainly so you can leave it alone.

Stacked imbalances (several in a row at consecutive prices) are the more meaningful pattern. They mark a price region that one side moved through with real aggression, and those regions frequently matter again later, because that is where somebody's inventory was built in a hurry.

The ratio is a setting, not a fact. Raise it and you get fewer, higher-quality flags; lower it and you get a chart covered in them. Choose one, write it down, and stop changing it — a threshold tuned until the chart looks right is a threshold that describes the past. The same applies to the comparison method itself: if your platform offers both, pick diagonal and leave it.

Absorption and exhaustion

These are the two patterns that justify the whole exercise, because neither is visible on any other kind of chart.

Absorption: heavy aggressive volume hitting one side, and price does not move. Sellers are throwing size at the bid and the bid holds. Somebody large is sitting there taking the other side with limit orders. When that aggression finally exhausts itself, the reversal can be sharp, because the aggressive side is now trapped and the passive side owns the inventory.

Exhaustion: the tail end of a move where aggressive volume in the trend direction collapses while price is still making marginal new extremes. The push is running on fumes.

Both are readings of a situation, not entries. The reason they are worth learning is that they answer a question no indicator can: is this level holding because nobody is testing it, or because somebody is defending it? Those are very different levels, and only order flow separates them.

What order flow cannot do

It cannot see resting size reliably. The depth of market shows intent to trade, and intent evaporates. Displayed liquidity is pulled constantly and legally, and iceberg orders deliberately hide real size. A wall in the book is not a fact about the future.

Spoofing exists and is difficult to distinguish live. It is illegal and prosecuted, and it still happens. A book read naively is a book that can be shown things.

It is not predictive, it is diagnostic. Order flow tells you the character of what just happened at a price. It has no opinion about the next hour. Traders who load a footprint expecting entries usually conclude the tool is broken; the tool is fine, the expectation was wrong.

The classification itself is an approximation. Bid/ask assignment depends on your data feed's timestamps and rules. Different platforms produce measurably different deltas from the same session. Treat the number as a good estimate, not a measurement.

It requires real, centralised volume. On futures, cleared through one exchange, the data is complete. On spot forex there is no consolidated tape at all, so “order flow” there is your broker's own flow — a sample of unknown size. On fragmented equity markets your feed sees part of the picture.

Low volume makes it lie. Overnight, in holiday sessions, or in an illiquid contract, delta and imbalance are dominated by a handful of orders. The same reading means completely different things at 15:30 and at 03:00.

It will not fix your exits. The most common reason a trader with good order-flow reading still loses is that the entries were never the problem. If you are not sure whether that is you, the trade analyzer settles it from your own log in about thirty seconds.

How to learn it without burning an account

Start with one instrument and one timeframe. Order flow reading is pattern recognition against a specific market's normal behaviour. Two thousand contracts is a big print in one market and a rounding error in another. You cannot build that calibration across five instruments at once.

Fix your settings before you look for setups. Cluster step, imbalance ratio, session boundaries. Every one of these can be tuned until the past looks obvious.

Read it at levels, not everywhere. A footprint in the middle of nowhere is noise with excellent resolution. The reason to open one is that price has arrived somewhere structurally interesting and you want to know whether the reaction is real. Which means you need the structure first — order flow is the second question, never the first.

Log the reads, not just the trades. Write down what you thought the flow said and what happened. That log is the only thing that converts “I watched a lot of footprints” into calibration.

Where to go next

The layer above this one — where volume has been done across the whole session rather than inside a single bar — is the volume profile guide. The independent third read, from the options market, is options positioning. Terms in plain language live in the glossary.