Volume Profile in Pine Script: Data, Bins, and Validation
A volume profile groups volume by price over a chosen window. In a custom Pine Script implementation, the result depends on the source data, row boundaries, and how bar volume is allocated. Defining those choices is more useful than presenting a profile as exact trade-by-trade information or guaranteed support and resistance.
Read the levels as a distribution
The point of control (POC) is the highest-volume price row. Value-area high and low bound a selected share of the distribution, often using a 70% setting. A wider row can combine nearby activity that a narrower row separates, so the POC may change when the resolution changes.
These levels summarize the selected history. They do not reveal a participant's identity, remaining inventory, or future willingness to trade. Treat a revisit or breakout as a hypothesis with a defined outcome, not a reaction that must occur.
TradingView's volume-profile guide explains its use of lower-timeframe data and directional up/down volume. Its volume type depends on the instrument and feed; a price-update count is not the same unit as exchange-traded shares or contracts.
Specify the calculation before choosing a data structure
| Decision | What to document |
|---|---|
| Window | Fixed timestamps, rolling bars, visible range, or named session |
| Data | Symbol, lower timeframe, volume type, and available history |
| Rows | Price origin, tick-aligned width, and boundary convention |
| Allocation | How one source bar's volume is distributed among rows |
| POC | Tie-breaking when multiple rows have equal volume |
| Value area | Target percentage and row-expansion method |
| Updates | Developing profile or a frozen completed profile |
request.security_lower_tf() returns arrays of lower-timeframe values. It does not recover every transaction from ordinary intrabar OHLCV data. If a one-minute bar spans several price rows, its OHLCV values alone do not tell you the exact volume traded at each price inside that bar. See the Pine data-request documentation.
A close-bin method assigns the entire bar's volume to its closing-price row. A range-allocation method distributes it under a stated assumption. These are different approximations, and neither should be labeled exact without supporting data. State which one your implementation uses.
Maps and arrays store accumulated values; changing the container does not create more granular source data. Compare performance with the Pine Profiler on the actual implementation before claiming that one structure is faster.
Use conservation checks
For a hypothetical three-row example, suppose the allocated volumes are 100, 300, and 200 units. The total is 600 and the middle row is the POC. Those values illustrate an arithmetic check, not a market result.
For each window, compare the sum of row allocations with the source volume included in that window. Investigate discrepancies caused by dropped data, overlapping boundaries, or double-counted intrabars. Define what happens when a source bar crosses the start or end of your selected period.
A value-area calculation needs more than a percentage label. Different expansion and tie-breaking rules can produce different boundaries on the same distribution. Record the algorithm and test it on a small hand-checkable example before interpreting a chart.
Sessions are a choice, not contamination
A regular-session profile and a full-session profile answer different questions. Overnight activity is not inherently bad data. Specify the session, exchange timezone, and reset condition, then verify the first and last included bars.
When a strategy uses yesterday's completed profile, freeze those levels at the correct boundary. Do not calculate a full day's final distribution and then apply it to decisions made earlier that same day. A developing profile legitimately changes as new volume arrives; a frozen historical reference should not silently absorb later data.
Compare with a built-in profile carefully
Align the symbol, session, time window, row settings, and chart type first. If your custom implementation uses a different lower timeframe or allocation method, a mismatch can be expected rather than evidence of a bug in either tool. Nonstandard charts can alter the price and volume representation; standard candles simplify an initial comparison.
Inspect totals and a few rows before comparing the POC line. A matching POC alone is a weak test because many different distributions can have the same maximum row.
From level to strategy
A condition such as "price returns to the prior profile POC" still needs a tolerance, confirmation time, entry rule, exit rule, and invalidation. Specify whether the reference is a row boundary or midpoint. Then add sizing, costs, and fill assumptions to a test.
This article reports no NVDA, TSLA, or AMD trades and no proven increase in signal accuracy. The practical goal is a reproducible profile whose data and timing can be audited before any trading interpretation is evaluated.