Auction market theory

Steidlmayer, the Chicago Board of Trade, and the auction room

The framework has a name and a date, which is already rare in this trade. J. Peter Steidlmayer, a floor trader at the Chicago Board of Trade, worked out in the early 1980s a way of picturing a session, which the exchange itself began distributing in the middle of the decade under the name Market Profile. His 1986 book, Markets and Market Logic, written with Kevin Koy, fixes the vocabulary that follows.

The founding idea fits in one sentence, and it runs against instinct: a market's job is neither to rise nor to fall, it is to find a price where business gets done. An exchange behaves like an auction room: it advertises a price, watches who answers, and if nobody answers it advertises another one. The session closes, the search starts over the next day, and that close is what makes a session measurable: it has a beginning, an end, and a distribution of its own.

The mechanism is a two way auction. Price rises to go looking for sellers, until it gets high enough to put buyers off. It falls again to go looking for buyers, until it gets low enough to put sellers off. Between those two refusals sits the range both sides will do business in.

One observation follows, checkable on any chart: a market spends the bulk of its time inside a handful of areas narrow in price and thick in volume, linked by moves that are wide in price and nearly empty. This is not a claim to be believed, it is what the measurement returns.

Value, in the sense this framework gives it

The word value is a trap, because it means something else elsewhere. To a financial analyst, the value of a share is what it ought to be worth given its earnings. That meaning plays no part here. Value is purely descriptive: the price range both sides accepted, measured by the volume that traded inside it. People can argue for ever about what an asset ought to be worth, not about the price at which the most contracts changed hands yesterday.

This value is not fixed. It shifts, session after session, and the shift is what carries the useful information. A value that rebuilds itself higher three days running says agreement is forming higher up. A value that stays where it was says nothing has changed, however busy the candles inside it look.

Three relative positions are enough to describe a session in progress. Price is inside yesterday's value, and the market is in balance. It is outside value and does not come back, so acceptance is being built somewhere else. It is outside value and drops straight back in, so a price was tried, refused, and yesterday's value still stands.

⚠️ Nothing in those three positions tells you what to do. They tell you where you are, which is a great deal and is not the same thing. Mistaking a description of state for a reason to enter is the first way to lose money with this lens.

How a distribution of volume by price gets built

An ordinary chart shows price against time. A distribution by price turns the question ninety degrees: for each price, it stacks up everything that traded there during the period, and it could not care less about the order in which any of it happened. The result reads sideways, in horizontal bars.

Two ways of counting coexist, and confusing them keeps a lot of misunderstandings alive. The original version counts time: the session is cut into thirty minute brackets, each bracket is given a letter, written in at every price the market touched during the bracket. That is the TPO, for time price opportunity. Your software today draws instead the volume actually exchanged at each price. The two produce close shapes without being interchangeable.

The shape you get tends towards a bell: plenty of activity in the middle, less and less as you move away from it. This is not some elegant statistical accident, it is the signature of the auction. The middle is where both sides found business to do, and the extremes are the prices refused by one side or the other.

Some sessions produce no bell at all, but two humps split by a hollow, or a stretched distribution with no readable centre. Those are often the instructive ones: an irregular shape flags a market that changed its mind halfway through, or one that did not finish the work it had started.

The point of control and the value area

The point of control is the price at which the most volume traded, in other words the longest bar of the distribution. It is the price the market agreed on most during the period observed. It is neither support nor resistance, and presenting it as one throws away what it offers: it is the centre of gravity of the trading.

The value area is the range holding roughly 70% of the period's activity. You build it by starting at the point of control and adding the busiest neighbouring prices until you reach the threshold. The figure of 70% has nothing magic about it: it is the ground one standard deviation would cover if the distribution were normal, which it never quite is. The threshold stays a convention, kept because it yields a readable range, and most software lets you change it.

What those two markers give you is modest and solid. They give you a shared vocabulary, which describes a whole session in three numbers instead of an impression. They give you addresses above all: the value area high, the value area low and the point of control are three prices to write down before the open and find again the following day.

⚠️ A point of control is not more reliable because its bar looks long on screen. A quiet session produces a very sharp one of little consequence, because everything happened in the same spot. A heavily traded session produces a wide distribution where the same marker carries far more weight. The length of a bar is read against the total volume of the period, never on its own.

What time spent at a price tells you

A candlestick chart encodes four numbers per interval, the open, the close and the two extremes. Time shows up only as one more slot on the axis, and every candle takes the same width whether one contract or a hundred thousand traded inside it. A candle therefore says what happened during one interval, never how many intervals the market spent in the same place. Counting time by price instead of price by interval is the one contribution this framework shares with no other.

An area price crossed fast and an area price lived in are not the same object, even when they take up exactly the same height on your screen. In the first, almost nobody had time to make a decision. In the second, positions were opened, closed, added to and regretted all session long.

That difference has a mechanical consequence rather than a psychological one. One confusion creeps in here and is worth clearing at once: volume traded is not the number of positions left open, since most of a large volume is rotation, opened and closed within the day. The more people go through a price, the more live positions sit on it all the same. When price comes back, it meets people wanting out at breakeven, others wanting to add, others who had been waiting for that level to sell. Where nobody traded, it meets none of that.

Reversing that reading is more useful still. Price cutting in ten minutes through an area where it had spent three days says yesterday's agreement no longer exists. Price settling into an area it had once crossed in a rush says agreement is forming where there was none. Those tell you about the regime, not about a trade, and they go with the market structure lesson, which describes the same switch in the vocabulary of highs and lows. No tool is needed to start: places where candle bodies overlap for a long stretch have been lived in.

Balance, imbalance, and range extension

A market has only two states in this framework, and it alternates between them without ever stopping. In balance, both sides agree on the range, price rotates inside it, the distribution thickens and time passes. In imbalance, one side wins, price leaves the range and covers a lot of ground in very few transactions.

Steidlmayer had a simple marker for spotting the switch while it happens: the initial balance, meaning the range of the first hour of trading, which is the first two thirty minute brackets of the lettering seen earlier. Anything that trades beyond it is a range extension. A session without extension is a balance session, where the first hour was enough to frame the whole day. A large extension on one side only signals imbalance instead.

The marker has aged on one point, and saying so beats copying it out unchanged. It was designed for markets with a short session, a sharp open and a trading floor. On an instrument quoted around the clock, the first hour needs redefining, usually around the open of its busiest session. The logic does not age: compare the range already covered with the one laid down at the start.

The full sequence reads at every scale. An area of balance gets built, it eventually stops suiting one of the two sides, price leaves it in imbalance, covers some distance, then builds a new area of balance higher or lower. A market is never anywhere other than in one of those two phases.

Low volume areas, and why some prices act as magnets

A distribution does not only show its humps, it shows its hollows too, and those give the most usable markers in the lesson. A low volume area, inside one distribution or between two neighbouring ones, is a place where next to nothing traded. Nobody agreed there. When price comes back, it goes through fast, upwards as readily as downwards, and that speed repeats often enough to be worth counting on.

Look closely at what that hollow measures, because this is where people get it wrong. It does not say the order book is empty there right now: a book rebuilds itself continuously and there are orders at every price. It says nobody built a position there in the past, so nobody holds inventory to defend and nobody has a reason to stop. That is an absence of interest, not an absence of orders, and the one thing it lets you anticipate is a speed of passage. Direction stays entirely open.

The corollary is the sentence to take away from this lesson: where business was done easily, it gets done easily again. A volume cluster left behind acts as a magnet, not because it attracts anything, but because a participant who has to trade in size picks the place where counterparty can be found, and that place is the one where there already was some. A point of control left untouched is therefore a serious candidate for a return, at a frequency that depends on the instrument and that the exercise at the end of this lesson will have you measure on yours rather than take on trust.

⚠️ A magnet is not a guarantee. A market can perfectly well settle into an empty area and build a brand new distribution there, which is exactly how fresh value gets formed. The right way to use it is therefore not to bet on the crossing, it is to know before you enter whether the area you are aiming at is empty or full.

Using this lens honestly

A profile triggers nothing. It has no entry signal, no condition, no trade direction, and anything sold to you as such is a layer somebody else added on top. What the lens gives you is rarer: it helps you choose where, once you have decided what to do for other reasons.

Where to put a target, first. A target set on a volume cluster left behind is a reasonable thing to ask of the market, because it is a place the market has already shown it can trade in size. A target set in the middle of a void is arbitrary: nothing holds price at that exact spot rather than at the next one along.

Where to put a stop, next, and this is where the lens repays what it cost you to learn. A stop placed inside a volume cluster will be hit by ordinary rotation, since rotating is exactly what price does inside a cluster. The same stop placed behind the area survives that rotation and only triggers if the market has genuinely left balance, at the cost of a smaller position, worked out the way the risk lesson showed you.

⛔ One warning governs all the rest: this reading requires reported volume, meaning a market where every execution is published. That is not the same thing as a single order book, a share trades on several venues at once and its volume is no less real for it, and the order flow lesson sorts this out instrument by instrument. Take the consequence from here: wherever your platform counts price changes instead of a volume, you are reading relative activity and you cannot build a value area. The TPO version still applies everywhere, which is no accident: the original framework counted time because volume by price was not published intraday back then.

Practice: reading one session's profile

The exercise takes twenty minutes and asks for nothing more than charting software able to display a volume profile, which free tools do these days. Pick an instrument whose volume is reported and a session that has already finished: you are not trying to predict anything, you are describing something that is over.

Display the previous session's profile. Write three numbers on a sheet of paper, the point of control, the value area high and the value area low. Then write in one line the shape you can see: a regular bell, two humps split by a hollow, or a stretched profile with no readable centre.

Now open the following session and answer three questions, in this order. Did price open inside yesterday's value area, above it or below it? If it opened outside, did it come back, and how long did that take? Was yesterday's point of control touched during the session, yes or no?

Do it again over ten consecutive sessions, on the same instrument, in a table of seven columns: the date, the three numbers, the shape, where the open fell, and whether the point of control was touched. You will then hold something no course can hand you: the frequency, on your own instrument, at which yesterday's point of control gets revisited. If it is poor, you have just learnt that this lens will not do much for you there, for twenty minutes of work.