Understanding order flow

The most repeated line in trading is impossible

“Price rises because there are more buyers than sellers.” You have read it a hundred times, it opens serious textbooks, and it describes nothing at all. A market is not a poll, it is a matching machine. A transaction is one buyer and one seller, for the same quantity, in the same instant, at the same price. The count on one side always equals the count on the other, at the microsecond as much as over the year.

What the line means, when whoever says it has thought about it, is that there was more willingness to buy than to sell. The trouble is that the lazy version sends you looking for a crowd, while the market counts nobody. It suggests that somewhere there is a scoreboard of buyers against sellers that a good tool would display. That scoreboard does not exist. The indicators that claim to show it, the delta chief among them, measure a different thing that has a precise name: they file every transaction on the side of whoever crossed the spread to get it. That is a count of aggression, not a count of participants, and the distinction carries the whole lesson.

The exact formulation fits in one line, and everything else in this lesson follows from it: price rises when nothing is left to sell at the current price. Buyers were not more numerous, sellers were exhausted. The quantity traded stays perfectly symmetrical on both sides. What is asymmetrical is the urgency: one camp agrees to wait, the other insists on being served immediately, and that difference is the only thing that moves a price.

The mechanism has a name, the continuous double auction, and it describes a stock exchange just as well as a futures market. Two queues of orders face each other, one offering to buy, one to sell, and nothing happens until somebody crosses the gap between them. Everything that follows in this lesson comes down to watching who crosses, what it cost them, and what was standing in the way.

What moves price: impatience

There are two ways of being in the book, and only one of them moves a price. A limit order resting away from the market is patient: you pick your price, your order joins the queue at that price, and it sits there until somebody comes and fills it. The market order is impatient: you pick no price, you take whatever is available, at whatever it costs. The first waits and displays, the second crosses and consumes. The line does not in fact fall between the two names, and the confusion is a stubborn one: a buy limit placed above the best offer fills on the spot and hits the book exactly as a market order would, save that it refuses to pay beyond its limit. An order is patient because it waits in the queue, never because the word limit appears in its name.

The queue is not a heap, it is ordered by a simple rule, price then time priority: the better price goes first, and at equal price, whoever arrived first is filled first. That is the rule on the great majority of books without being a law of nature: some futures contracts allocate pro rata to order size, and there the order of arrival counts for nothing. Where time does count, a patient order placed early at a good price has real value, and the same order placed at the last moment behind a queue of two thousand contracts has almost none.

Albert Kyle, in 1985, in the paper that founded modern market microstructure, Continuous Auctions and Insider Trading, formalised the result still in use today: the price move is proportional to the aggressive order flow, and the coefficient of proportionality measures exactly the inverse of available depth. A thick book takes a great deal of aggression for very little movement, a thin book does the opposite. The same order, sent two hours apart, therefore does not cost the same.

The same year, Lawrence Glosten and Paul Milgrom explained why the gap between the best bid and the best offer is never zero: whoever waits patiently in the book runs the risk of filling somebody better informed, and the gap is the price of that risk. That is why it widens on economic releases and during quiet hours, at exactly the moments when information is most dangerous and counterparties are scarcest.

The practical consequence concerns you at every click. When you buy at market, you are the impatient one: you pay the spread, you consume what was offered, and you nudge the price a little. When you place a limit order you save the spread and accept a different risk, that of never being filled and watching the move leave without you. Neither choice is free: the first pays the spread for certain, the second pays in missed opportunities, and that second cost shows up on no broker statement.

One level runs out, and the next one takes over

A price level is not a line, it is a finite stack of quantity. At a given price there is a certain number of contracts or shares on offer, not one more. When an impatient buyer takes them all, nothing is left at that price: the next transaction necessarily happens one step higher, because that is the first place where something is still available to buy. Price does not decide to rise, it notices it has no choice left.

A six-point rise is therefore not an event, it is a chain of successive exhaustions. That changes the reading, because the links of the chain are not equal. Three thin levels empty in two seconds, and the fourth one, where somebody posted ten times the quantity, stops everything. What your chart will later call a resistance is very often nothing more than that fourth level.

This reading also explains speed: price crosses at full pace the areas where nobody posted anything, and grinds through those where a lot of quantity is waiting. The auction theory lesson takes it from there, at the scale of time spent somewhere, and there is no reason to redo it here.

⚠️ The wall is not a monument. A book rebuilds itself constantly, and nothing obliges the hundred and thirty contracts in the example to still be there ten minutes later. What you are looking at is a snapshot, not a property of the level, which is why a resistance sometimes gives way with no fight at all: it was not forced, it stopped being stocked.

One consequence reads straight off a candlestick chart, with no tool at all. A long-bodied candle with ordinary volume crossed emptiness. A small-bodied candle with enormous volume met a wall. In a table of session statistics the two are indistinguishable, and they are not telling the same story.

Absorption or a void: the ratio that sums it all up

Once you accept that price advances by eating substance, reading flow reduces to one ratio: how much movement do you get for how much volume traded? That ratio cannot be read on a candle, since a candle only shows the numerator. You need the denominator, and the denominator is volume.

Richard Wyckoff, a New York broker and publisher of the early twentieth century, was already teaching this reading under the name law of effort and result. Effort is the volume traded. Result is the ground covered by price. When the two are proportionate the move is ordinary and teaches you nothing. When they diverge there is information, and it is essentially the only information order flow provides.

A fast move on light volume signals a void: few resting orders were there, price slid without consuming anything. The screen is spectacular, there is no substance behind it. Nothing was eaten, so nothing will stop price coming back the same way, just as fast, in the other direction.

A slow move on heavy volume signals absorption: many orders were consumed for very little ground. The screen is laborious, the substance is real. Somebody on the other side kept selling as the buying came in, and that supply has left the book. It will not be waiting there next time price comes back.

The point that costs beginners the most fits in one sentence: those two moves display identically on a candlestick chart. Same body, same wick, same colour. Everything that separates them lives in the pane underneath, the one half the traders switch off because it “takes up room”.

What each of them announces

A void leaves an area that stays open in both directions. Price went through without consuming anything, so nothing was built there, and it can go back through in the other direction at the same speed. That is the deep reason why some impressive rises are handed back in full within a single session, with no bad news arriving in between.

Absorption announces the opposite: a place where a great deal of supply was eaten. When impatient buyers exhausted thousands of contracts offered at one price, those particular contracts are gone, and the ceiling that was holding price down has been paid for once. The next attempt to go up will not have to pay for that one again. That is all the subtraction entitles you to say, and it is already useful.

⛔ Absorption on the buy side does not build a floor, whatever you read to the contrary. A floor is made of resting buy orders below the price, and eating supply above puts down not one of them: it removes an obstacle to going up, which is the exact opposite of support from underneath. A book has no memory either, it holds nothing at any instant but the live orders, and nothing stops fresh sellers from settling at the same price a minute later. If the level holds on the way back, it is because somebody put buyers there, not because sellers were once cleared out of it.

⚠️ Neither is a signal, and the nuance is expensive. Heavy volume with no progress can equally mean that the aggressors are failing: buyers hammering a wall and getting nowhere are exhausting themselves, not absorbing. To settle it, always look at which way the result points. Plenty of effort and no progress upward means the other camp is winning.

The honest formulation, the one a beginner can hold without telling themselves a story, is very plain: this move was expensive, or this move cost nothing. Nothing more than that. Nobody can know who was buying or why, and the narratives written about it are reconstructions produced after the fact.

Volume answers one question, and only one

Volume does not say whether it is bullish. It does not say who bought. It does not say what happens next. It answers one question and one only: how much had to be traded to obtain this movement? Nor does it split into buy volume and sell volume, since every contract traded is bought by somebody and sold by somebody: the green and red columns certain tools offer file each transaction on the side of whoever crossed the spread, which is again a measure of aggression and not a division of the volume.

On a break, the answer becomes blunt. A break carried by volume well above the ordinary says the level was genuinely contested: there was substance there, and somebody took it. The same break with no volume mostly says nobody was there, which is the signature of the false break, the one that slides back into the range two candles later having proved nothing at all.

Volume can only ever be compared with itself, and that precaution matters more than the rest. Compare a candle with the last twenty candles of the same timeframe and the same hour, never with a daily average. The open and the close of major sessions carry structurally more volume than mid-afternoon, and what passes for heavy volume at lunchtime can be a trivial print at the open.

⛔ This reading needs REAL volume. The lesson “Getting started with order flow” covers which markets have it, and why the number shown on a spot currency pair is not one. What has to be added here is a distinction that even real-volume markets impose: centralised does not mean single.

A futures contract trades on one book, and its volume is a total with no ambiguity. A listed share is another matter: in the United States as in Europe, the same share changes hands in parallel across several exchanges, alternative platforms and internalising desks. The number you read then depends on which execution venues your source aggregates, and two platforms routinely display two different volumes for the same share in the same hour. Before building a rule on a volume threshold, find out where the number comes from: a threshold calibrated on one source and applied to another never fires at the same moments.

The order book, and what it does not tell you

The order book is the list of resting orders at each price, on both sides, with their quantities. On futures markets it is also called depth of market. It is the view closest to the real mechanism, and it is also the most deceptive one for a newcomer, because it shows intentions rather than facts.

First trap: a displayed quantity can vanish. A resting order is cancelled in microseconds, and a wall of five hundred contracts that evaporates the instant price approaches was never meant to be filled. Deliberately deceptive display has a name, spoofing, and it is illegal on regulated markets: the United States banned it explicitly in the Dodd-Frank Act of 2010, and convictions have followed. Banned does not mean gone, and cancelling an order remains perfectly legitimate in the overwhelming majority of cases.

Second trap, exactly symmetrical: a quantity can be hidden. An iceberg order displays only a fraction of its size and replenishes automatically as it gets filled. The book then understates what is really on the other side. Its signature is recognisable: a level that refills every time it is hit, and refuses to give way long after it should have been emptied.

Hence the rule worth keeping even for someone who will never open a book in their life: a book is read through what executes, not through what is displayed. The professionals who use one read the trade tape alongside it, meaning the running list of what actually changed hands, price by price. The display is a statement of intent. The execution is a fact.

⚠️ The book is a very short-horizon tool, on centralised markets, and it is a discipline of its own. For someone deciding on a four-hour chart it adds nothing but extra stress and a permanent temptation to get out too early. Volume by price is more than enough, and it takes a second to read.

Slippage, and where it comes from

Slippage is the gap between the price you were looking at when you clicked and the average price you actually got. Plenty of beginners read it as broker dishonesty. In the overwhelming majority of cases it is simply the thickness of the book making itself felt, and the calculation can be checked by hand.

Your market order is filled level by level. If you ask for a hundred lots and the first level only offers thirty, you take the thirty, then you go up for the next ones. The price you pay is the weighted average of the pieces, not the price displayed before your click. That displayed price was only ever valid for the quantity offered at it, which nobody writes next to the button.

Three circumstances amplify the effect, and they are unfortunately the three moments when the urge to act is strongest. Quiet hours and thinly traded instruments, where the book is naturally thin. Economic releases, where patient orders step away a few seconds before the announcement, for exactly the reason Glosten and Milgrom described. Stop cascades, finally, which throw a mass of impatient orders at the market precisely when the least substance is left on the other side.

⛔ A plain stop order does not guarantee a price, it guarantees an attempt to exit: once triggered it becomes a market order and takes what it finds. A stop limit guarantees the reverse, a price but not the exit, and it can leave you holding a position you believed was closed. Choosing between the two is a decision, not a default setting, and it is taken before the position is opened.

The consequence for your own records is concrete. Slippage is a cost, exactly like the spread and the commission, and it belongs in your journal. A statistical edge measured without it is a theoretical edge: this is precisely the kind of cost that turns a method profitable on paper into a neutral one in the real world, without any single trade ever looking abnormal.

Practise: ten sessions, three columns

Three ideas from this lesson travel everywhere, including to somebody who will never open an order book. A price rises by exhaustion, not by counting a crowd. The ratio between movement and volume separates a fragile move from a solid one. What is displayed is not what executes, which holds for a book and for very nearly everything else in this business.

Here is the exercise that anchors them, and it fits in ten lines on a sheet of paper. Pick an instrument with real volume, a futures contract or a liquid share. Take the last ten sessions and write three columns: the date, the range of the session in points, meaning the high minus the low, and the total volume of the session. Nothing else, no interpretation.

Then rank the ten sessions twice: once by decreasing range, once by decreasing volume. Compare the two rankings. They broadly resemble each other, which is normal, since a market that moves trades more. What matters to you are the disagreements: a large range in the bottom third of volumes is a void, a small range in the top third of volumes is absorption.

Second pass, the one that does the work. For every disagreement you spotted, look at what the two or three following sessions did. Write it down in one sentence, without trying to conclude anything. Ten sessions are not a statistic and this exercise is not a test of a method: its purpose is to fuse, in your eye, the shape of a candle and the height of a volume bar, until you cannot look at one without seeing the other.

Keep the sheet. Crossed with your own trades for the month, it answers a question very few beginners know how to ask: am I losing mainly on the days when the movement cost nothing? What comes out is not an answer, it is a lead drawn from your own data rather than from an opinion. Ten sessions will not settle it: they are enough to tell you whether the question deserves asking again over a hundred.