Volume and the volume profile

Bar volume is how many coins (or dollars) went through in trades during that bar. Each trade is counted once: if a buyer took 2 BTC and a seller gave the same 2 BTC, volume grew by 2 BTC, not 4. Hence the main property of volume: in every trade what was bought equals what was sold, and volume by itself does not say who won. It says how actively people traded.

Volume comes in coins and in dollars. In coins — how many BTC (ETH, SOL…) changed hands; in dollars — the sum of price × quantity over all trades. With a noticeable price move inside the bar, dollars cannot be obtained exactly by multiplying coins by the close: each trade went at its own price. Volumes of different coins can only be compared in dollars: 1,000 of a cheap coin and 1,000 BTC are incomparable sums.

Illustrative example. During a five-minute bar 120 BTC traded at an average price of 61,500. The dollar volume is 120 × 61,500 = 7,380,000, about $7.4M. If the price went from 61,000 to 62,000 during the bar and we multiply 120 by the 62,000 close, we get $7.44M — a $60K error, under one percent. On a calm bar that is a trifle; on a sharp bar of several percent it is not.

Now three pictures that volume helps tell apart. High volume on a level breakout: many fought over the level, there were many trades on both sides. High volume without price movement: market orders kept coming and the price stood still — so they were met by limit orders (this may be a wall from section 4). Low volume on a move: almost nothing stood in the price's way, the book was thin; the move went “through empty space”.

One coin trades on a dozen venues at once, and each has its own volume. One exchange's volume is not the whole market. Illustrative example: in an hour of bitcoin, Binance traded 1,800 BTC, Bybit 900, OKX 600, Coinbase spot 300. 3,600 BTC in all, and Binance is exactly half. A service showing only Binance shows “1,800”; another that adds up four venues shows “3,600”, and both are right, each for its own set.

Hence the rule: volume from different sources can only be compared if the set of exchanges is the same. Within one source a change of set matters too: if a venue drops out for an hour, total volume falls although the market did not calm down. A good source names which venues went into the number.

Two more differences. Spot and derivatives are different markets: on perpetual futures for large coins volume is usually several times that of spot, because there is leverage and no need to own the coin. And venue volumes differ in quality: large long-established exchanges and venues where volume is inflated by trading with oneself must not be mixed in one sum. So it matters not only “how much” but also “where”.

How this works in EVA. Chart candles come from one exchange and their volume is counted in coins. Everything EVA builds from its own trades — the volume profile of this section and the by-price breakdown of a candle from section 9 — is counted over all the feeds the terminal collects, spot together with derivatives, and in dollars. So a candle's volume and the profile's total for the same stretch do not match and should not: they are different sets of exchanges and different units. They cannot be cross-checked directly — only each against itself.

Bar volume answers “how much was traded during this time”. The volume profile answers a different question: “how much was traded at each price” over a chosen stretch — an hour, a day, a week. It is a histogram turned on its side: price vertically, bar length is the volume at that price. Time disappears from the profile: it does not matter when trading happened at 83,300, only how much.

In EVA this is the “Volumes” layer on the chart. The “Volume profile” view is the classic one: a bar at each price, made of buys and sells. The “Profile by side” view is the same volume split out: sells on the left, buys on the right. The stretch is set with the mouse (its edges can be moved later, there can be several stretches) or by ready periods: the day from midnight UTC, the week, the Tokyo, London or New York session.

POC (point of control) is the price with the most volume in the stretch, the longest bar. Not the close, not the middle of the range and not the price with the most trades — precisely the one with the most volume.

The value area is a continuous price range around the POC where about 70% of the stretch's volume traded. Its upper edge is VAH (value area high), the lower one VAL (value area low). In EVA the area is built like this: start at the POC and at each step add the neighbouring bar — above or below — that is larger, until 70% of the volume is reached. So the area always contains the POC and is always continuous, with no gaps. Building stops at the first bar that crosses 70%, and may overshoot noticeably; the actual share is written next to the edges.

The area need not be symmetric around the POC. If volume leans one way, the area grows that way, and the POC can end up at its very edge — even coincide with VAL or VAH.

Illustrative example (in the figure below). A stretch with 100 units of volume in all: 100 — 4, 101 — 10, 102 — 18, 103 — 30, 104 — 22, 105 — 9, 106 — 7. The POC is 103 (30). We need to reach 70.

• Start: 103, 30 reached.
• Step 1: neighbours 102 (18) and 104 (22). 104 is larger — take it: 30 + 22 = 52.
• Step 2: the neighbours are now 102 (18) and 105 (9). Take 102: 52 + 18 = 70. 70% reached — stop.

The value area is 102–104: VAL = 102, VAH = 104, share 70%. Note: at step 2 the upper neighbour is no longer 104 (it is in the area) but 105. A typical mistake is to compare with the POC's original neighbours every time, or to build the area equally up and down.

The profile in EVA has two sources, and they differ. When the terminal has trades by price (it has them for the last hours in memory and deeper in the archive), the profile is real: it is known how much went through at each price and who was the aggressor, and the bar is coloured in two parts. When such data is missing, the profile is estimated from bar ranges: a candle's volume is spread over its range in proportion to the overlap with each price. Such a profile knows no sides and is not two-coloured, and inside a big candle accuracy drops: everything that actually traded near one edge ends up smeared across the whole candle. The bottom of the stretch always says which way it was built.

High volume without movement is a wide part of the profile where the price moved in a narrow range for a long time and there were many trades. Here market orders were met by limit orders: someone large was building or unloading a position, and the market “agreed” on a price. Who exactly won — buyers or sellers — the volume profile does not say: in every trade as much was bought as was sold. It can be accumulation before a rise, distribution before a fall, or just a calm day.

Thin spots of the profile are prices where little traded. The usual explanation is that the price passed quickly, nothing stood in its way. But there is a second, duller explanation: the price simply spent little time there. A profile is time multiplied by activity, and a thin stretch often means “briefly”, not “easily”. Below is an example from our archive where this shows in numbers.

Session profiles. EVA's ready periods are not only the day and the week but also the Tokyo, London and New York sessions (in each city's local time, with each switching to daylight saving on its own date). A session profile answers “where did these particular hours agree”: for example, Asian hours kept the POC at one price and New York moved the volume higher. Comparing neighbouring sessions' profiles is useful — you see whether the agreement shifted or the market traded in the same place.

But sessions are a calendar, not the market. Crypto trades around the clock, and “Tokyo has closed” neither forbids nor promises anything. The boundaries follow these cities' stock-market hours, not crypto exchange activity. A session profile is a convenient way to slice the day, not a special level.