Volume · Method
Volume Spread Analysis
A way of reading a single bar: how far price travelled, where it closed inside that range, and how much volume it took. The premise is that a large position cannot be built or unloaded without leaving a trace in those three numbers.
The premise
Every indicator elsewhere on this site compresses a history into one number per bar. Volume spread analysis does the opposite: it refuses the compression and reads the bar itself, on the argument that the three quantities a bar contains (the spread, the position of the close within it, and the volume behind it) carry information that any average of them destroys.
The reasoning behind it is mundane and hard to argue with. A position large enough to matter cannot be established or liquidated in one session, because the shares have to come from somewhere and there are only so many available at any price. The trace of that activity is volume that does not match the movement it produced, a great deal of trading with very little to show for it, or a substantial move that required almost none.
What follows from the premise is a set of named readings. They are descriptions of a session’s character rather than forecasts, and every one of them appears regularly in markets that then do nothing whatever.
The four readings that matter
Dozens of named bars circulate in the literature. Four account for most of what the method actually claims, and they come in two pairs, one about an advance that has run out of participants, one about a decline that has run out of sellers.
| Reading | The bar | The claim |
|---|---|---|
| No demand | Up bar, narrow spread, weak close, volume below the last two bars | The advance found no participation. Price rose because nothing was offered, not because anything was bought. |
| Upthrust | New high, wide spread, heavy volume, closes at or near the low | The break above the high was met by supply and rejected inside the session. |
| Stopping volume | Heavy volume in a decline, closes well above the low | Selling was absorbed by buying large enough to arrest it before the close. |
| No supply | Down bar, narrow spread, volume below the last two bars | The decline met no selling of consequence, the mirror of no demand. |
Each definition contains a judgement (narrow compared with what, light compared with what), and that is where the method’s weakness lives rather than in the logic. The comparison is conventionally against the two or three preceding bars rather than a long average, which keeps it local and makes it sensitive to exactly how the preceding move is delimited.
Where the method is weakest
| Situation | What goes wrong |
|---|---|
| Judgement in the definitions | "Narrow spread" and "light volume" are relative to a window the analyst chooses. Two people coding the same rules identify different bars. |
| Reading one bar | Every named bar occurs constantly in markets that go on to do nothing. Confirmation in the following sessions is not optional. |
| Index rebalance and expiry | Mechanical volume from funds obliged to trade produces textbook-looking bars that carry no opinion at all. |
| Illiquid instruments | A handful of trades sets both the spread and the volume, so the bar describes those trades rather than a market. |
| Intraday without adjustment | Volume clusters at the open and the close, so bars from different parts of the session are not comparable inputs. |
| Hindsight selection | Charts in the literature are chosen after the outcome is known. The bars that were followed by nothing are not in the book. |
Reading it honestly
The method has a genuine advantage and a genuine liability, and they are the same property: it uses more of what the bar contains than any indicator does, and it asks a person to weigh it. Three habits keep that from becoming self-deception.
Define the comparison before you look. Decide what "light volume" means (lower than the previous two bars, or below the twenty-bar average), and apply it consistently. Almost every disagreement about whether a bar qualifies is a disagreement about the window, and it can be settled in advance rather than argued about afterwards.
Require the following bars to agree. An upthrust followed by a session that makes a new high and holds it was not an upthrust in any useful sense. The reading is a hypothesis about what happened inside one session, and the next two or three sessions are the only available test of it.
Keep a record, including the failures. Note the date, the reading, and what followed. This is the only correction available for a method with judgement in its definitions, and it is the part the literature omits, a book of charts chosen after the fact will convince anyone of anything, and a notebook of your own bars will not.
What the numbers add
Reading bars and computing indicators are usually presented as rival approaches, which is a mistake in both directions. An indicator applies one rule identically to every bar and cannot be talked into anything; bar reading uses information the indicator throws away and can be talked into a great deal. Each covers the other’s weakness.
The practical combination is narrow. Use the bar reading to notice a session where effort and result diverged, and a mechanical measure, Volume RSI over the same window, or the up/down volume split, to establish whether the surrounding period supports that impression. Then look outside the instrument: if breadth was narrowing while your bars said an advance had stopped finding buyers, two independent inputs agree, which is as close to confirmation as any of this gets.
Frequently asked questions
What does "effort against result" mean?
Volume is the effort and the spread is the result. A session that trades twice its usual volume and closes almost unchanged has expended a great deal of effort for very little movement, which says someone large was on the other side absorbing it. The reverse — a wide spread on light volume — says the move met almost nothing, which is why it is treated as unreliable rather than as strength. The method consists of comparing those two quantities bar by bar and noticing when they disagree.
Is volume spread analysis the same as Wyckoff analysis?
It comes from the same tradition and is narrower. Richard Wyckoff’s work in the early twentieth century described market phases (accumulation, markup, distribution, markdown), and the behaviour of a large operator through them. Volume spread analysis, developed later by Tom Williams, keeps the premise that the trace of large activity is visible on the tape and concentrates it into readings of individual bars and small groups of them. One is a framework for a whole market cycle; the other is a way of reading three bars.
What is a no-demand bar?
An up bar with a narrow spread, closing weakly within it, on volume lower than the two bars before it. The reading is that the advance found no participation: price rose because nothing was offered rather than because anything was bought. On its own it means very little, quiet sessions happen constantly, and it acquires weight when it appears after a substantial advance and is followed by an inability to make further progress.
What is an upthrust?
A bar that makes a new high on a wide spread and heavy volume, then closes at or near its low. The interpretation is that the move above the previous high triggered buying and resting stop orders, and that supply met all of it; the high was reached and rejected within the session. It is the single most cited bar in this method, and it is also frequently just a volatile session, which is why practitioners insist on reading the bars that follow rather than acting on the bar itself.
What is stopping volume?
Very heavy volume during a decline, on a bar that closes well above its low. The reading is that selling was met by buying large enough to arrest it inside the session. It appears near lows more often than at random, which is the asymmetry that runs through every volume and breadth measure: forced selling is concentrated and visible in a way that accumulation is not, so the heaviest bars in a decline tend to come at its end.
Does any of this constitute a signal?
No, and the practitioners worth reading are explicit about it. Each reading is a statement about the character of one session (heavy effort met by absorption, an advance without participation), and the same bar appears frequently in markets that go on to do nothing at all. What makes a reading worth noting is confirmation in the following bars, and the honest way to use the method is to write down what you saw and check it later, including on the many occasions it led nowhere.
Can it be tested?
With difficulty, and that is a real criticism rather than a defence. The definitions involve judgement at every step: how narrow is a narrow spread, how light is light volume, which bars count as "the preceding move", so two analysts coding the same rules produce different bar sets. Coding a strict version is possible and is worth doing before trusting anything, because the strict version generally identifies far fewer bars than a chart read by eye and performs less impressively than the literature suggests.
Does it work on intraday charts?
The readings are defined on any bar interval, and the volume comparison is the problem. Intraday volume is shaped by the session: the first and last few minutes carry a disproportionate share of the day’s trading, so a bar at the open will look heavy and one at midday light for reasons that have nothing to do with participation. Comparing an intraday bar against the average of nearby bars in the same part of the session, rather than against a flat average, is the minimum adjustment, and it is the distortion every volume reading on this page depends on.
What data do I need?
High, low, close and volume per bar, which is what any ordinary price feed provides. The open is not required for most readings, and the method predates the widespread use of candlesticks in the American market, which is why the classic descriptions talk about the spread and the position of the close rather than about bodies and wicks. Adjusted prices matter as much here as anywhere: an unadjusted split leaves a bar whose spread is meaningless.