Volume Profile

What a volume profile is

Most charts print volume as a bar under each candle. That tells you how busy each period was. It does not tell you where, inside the price range, the trading happened. A volume profile answers the second question. It splits the price axis into buckets, adds up the volume that traded in each bucket over a chosen window, and draws the totals as horizontal bars.

Two features are read off it. The point of control (POC) is the bucket with the largest volume. The value area is the band of buckets around the POC that together hold a chosen share of the total volume, most often 70 percent. Thick regions are sometimes called high volume nodes and thin ones low volume nodes. These are names for shapes in the data, not predictions.

How it is calculated

There are four steps, and each one has a setting that changes the answer.

• Pick the window. One day, one week, or the last 500 bars. A different window gives a different profile.
• Pick the bucket size. A bucket might be one dollar, ten ticks, or a fraction of the average range. Smaller buckets give a jagged profile and larger ones a smooth one.
• Assign volume to buckets. The simple method puts each bar's whole volume in the bucket holding its typical price or its close. A finer method spreads the volume across every bucket the bar touched. Tick data lets you use the actual trade prices.
• Find the POC and the value area. The POC is the largest bucket. To build the value area, start at the POC and keep adding the larger neighbour until the running total reaches the target share.

A worked example with made-up numbers

Take 12 bars, each with a made-up typical price and a made-up volume. The first three are 100.4 with 300, 100.8 with 500 and 101.3 with 800. Use buckets one dollar wide, and put each bar's whole volume in the bucket that holds its price. The table adds up the result.

The total is 5,950 units. The 101.00 to 102.00 bucket holds 3,800, which is 63.9 percent, so it is the POC. The 70 percent target is 4165 units. Starting from the POC, the value area adds the larger neighbour each time and stops once it reaches 5,200 units, or 87.4 percent, across 100.00 to 102.00. The remaining 750 units sit outside it. Only 100 units traded at 103 or higher, so that region is a thin node in this sample.

The numbers are made up to show the sum. Nothing here describes a real market, and the same method on real data gives a different shape every time.

How it differs from the volume bar

A volume bar has time on the horizontal axis and volume on the vertical. A profile has price on the vertical axis and volume on the horizontal. The two use the same data, regrouped. In the example, the largest single volume bar is bar 5 with 900 units. The largest bucket is 3,800 units, because it pools 5 bars that traded at nearby prices. Neither view is more correct. They answer different questions: when was it busy, and where was it busy.

It also differs from the VWAP indicator. VWAP is one number, the volume weighted average price, and it moves as the session goes on. A profile keeps the whole distribution. A profile with two thick regions and a thin gap can have the same average as a single hump, and the average alone hides the difference.

What data it needs

• Real traded volume. On an exchange with a reported volume figure, the profile reflects actual trades. On markets without a single central tape, the volume you see is the volume at one venue, or a count of price ticks, and the profile describes that source only. Check what your data provider counts.
• Fine enough prices. Bar data forces a guess about where inside the bar the volume traded. Tick or trade data removes the guess. The ideas in order flow use that same finer data.
• A fixed rule for the window. A session profile that resets each day and a rolling profile over the last N bars are different tools. A bot must say which one it uses and compute it the same way in a test and live.

How a bot can misuse it

• Treating the POC as a magnet or a wall. It is the past price where most volume traded. The price can pass straight through it. A rule that buys at the POC and sells at the value area edge is a test to run, not a fact to assume.
• Settings that move the answer. Change the bucket size from 1.00 to 0.50 and the POC can jump to a different place. Tuning the bucket size and window until the backtest looks good is overfitting.
• Look-ahead in the window. If the profile for a day is computed from the whole day and a test uses it at the open, the test has seen the future. Build the profile only from bars that had closed at the time of the decision.
• Breakout rules built on thin nodes. A rule that expects a fast move through a low volume region is a form of breakout trading, and it faces the same false breakout problem. A move that stalls costs the spread and slippage on the way in.
• Stacking it on other indicators. Adding the profile to a Bollinger band rule or an EMA crossover adds settings. Each new setting is another chance to fit the past.

How to check a rule that uses it

Write the exact window, bucket size and value area share before you test. Test on data the settings never saw, as in how to backtest. Count the trades, take the average result per trade after costs, and size each position with a fixed risk rule, as the position size calculator does. Then change the bucket size by a modest amount and see whether the result survives. A rule that works only at one bucket size is likely fitted.

What it does not tell you

It does not say who bought and who sold, whether volume at a level was building a position or closing one, or what happens next. It says where volume traded in the window you chose, with the bucket size you chose. This page is education, not a signal.