How to Read Trading Volume: Relative Volume, Divergences and Open Interest

“Price rose on heavy volume, so fresh money must be buying” is one of the most repeated conclusions in chart analysis—and one of the easiest to misuse. Volume does measure activity, but it does not reveal a one-sided flow of cash. Every completed trade has both a buyer and a seller. More importantly, a share traded on an exchange, a futures contract, a quote update in spot foreign exchange and a reported transaction in the wholesale gold market are not equivalent observations.

This is Part 4 of the Financial Chart Basics series. Part 3 on support and resistance explained why volume can strengthen the evidence behind a breakout without replacing price acceptance or structural confirmation. This guide develops that idea in full: identify what the volume field actually contains, normalise it against comparable trading periods, and only then interpret price-volume relationships, divergences and futures open interest.

Before reading the bar, identify the dataset

A conventional volume bar records the number of shares, lots or contracts exchanged during a time interval. It measures completed activity, not purchases minus sales. Labels such as “buy volume” and “sell volume” normally refer to an aggressor-classification rule: a data provider infers which side crossed the spread by comparing the execution with bid and ask quotes or with the previous trade. That can be useful, but it remains a classification rather than an account-level cash ledger.

Five fields that appear similar on a chart answer very different questions:

The same share may change hands several times in one session. Volume divided by free float therefore measures turnover intensity, not the percentage of unique investors who reallocated once. Traded value can be better for judging capital capacity, but comparisons across securities still need adjustments for price level, currency, free float, venue coverage and session length.

“High volume” only has meaning against a comparable clock

Most markets have a persistent intraday volume curve. Equity activity tends to cluster near the open and close. Futures volume often accelerates around the opening of the related cash market, economic releases and settlement. FX quote activity is generally greater during the London–New York overlap than during a quiet Asian afternoon. A 3 p.m. bar can therefore exceed an 11 a.m. bar without signalling any new information.

For a completed bar, a reproducible same-time relative-volume measure is:

Same-time RVOL = current bar volume / median volume at the same time over N comparable sessions

For cumulative intraday activity, compare equal progress through the session:

Cumulative RVOL = volume so far today / median volume by the same time over N comparable sessions

The median is less sensitive than the mean to a handful of event days. It does not remove every design choice: N, half-days, holidays, day-of-week effects and futures roll stages must be specified in advance. An unfinished 30-minute bar cannot be compared with a set of completed 30-minute bars. Nor can the first ten minutes of today be divided by historical full-session volume without systematically understating activity.

Working rule: ask “relative to what?” before calling volume unusual. Without a defined session, comparable time slice and data source, “high volume” is usually a visual impression rather than a testable observation.

The four price-volume combinations are hypotheses, not forecasts

Price and relative volume are often placed into four quadrants. Each quadrant suggests questions; none supplies a direction by itself.

Price can travel quickly on low volume when a thin order book allows small aggressive orders to cross several levels. It can also remain nearly unchanged on enormous volume when large opposing interests transact in the same area. The first situation is about scarce liquidity; the second may be absorption or position transfer. Bar height alone cannot distinguish them.

Why a price-volume divergence is only an alert

A divergence usually means price posts a new high or low while raw volume, relative volume, turnover or a derived volume indicator does not confirm it. The condition shows that the market is advancing differently from the prior swing. It does not prove an immediate reversal. A mature trend can continue on modest volume when little opposing inventory is offered, while the normalisation of volume after a major event can create a declining pattern with no deterioration in the underlying trend.

A testable divergence rule must define four items before the outcome is known: the chosen volume measure; the two confirmed swings being compared; whether price and activity use extremes, closes or window statistics; and the number of bars after which the signal expires. Selecting the two most attractive peaks after viewing the entire chart introduces selection bias.

An alert becomes a structural warning only when price also loses a defined area, fails to hold new highs or lows, sees countertrend RVOL expand, or breaks a pre-specified sequence of swings. Conversely, continued closes in the new area, a lower-volume pullback and renewed expansion in the trend direction should downgrade the earlier divergence.

What volume adds to breakouts, retests and absorption

Rising RVOL at a boundary shows that the crossing attracted more participation than comparable periods. That is stronger evidence than an isolated move in exceptionally quiet conditions, but it still does not guarantee success. An economic release can expand volume, volatility, spreads and slippage at the same time. Large resting supply may absorb aggressive buyers above resistance and leave a failed spike. A low-volume break is not automatically false either: when offers above the old range are sparse, price can relocate with relatively few trades.

Instead of labelling one bar, examine a three-stage sequence:

1. Departure: did price leave the zone with a meaningful body and volatility-adjusted distance, and was relative activity abnormal?
2. Acceptance: did subsequent completed bars remain outside, even after event volume moved back towards normal?
3. Retest: did activity contract into the old boundary and then recover as price moved away in the breakout direction, with executable spreads?

“Heavy volume without further gains” is often called distribution, while “heavy volume without further losses” is called accumulation. Volume cannot identify account ownership, so the neutral description is that aggressive flow was absorbed by opposing liquidity. Whether that is a top, a bottom or two-way event hedging depends on location, trend, later closes and execution conditions.

Futures volume and open interest answer separate questions

Futures volume counts contracts traded during a period. Open interest counts contracts that still exist at the reporting point. Each open contract connects one long with one short but is counted once, not once per side. The opening and closing status of both counterparties determines the change:

Price rising with open interest therefore means the stock of outstanding contracts expanded during the advance; it does not show that all new positions were bullish, because every new long has a short counterparty. Price rising while open interest falls can include short covering, pre-expiry reductions or broad risk compression. None is sufficient evidence that the trend must end.

Contract rolls are a frequent source of false conclusions. Falling volume and open interest in the expiring contract, alongside increases in the next active month, may simply be position migration. A continuous series can also alter price and volume patterns through its stitching and back-adjustment method. Analysis should return to the tradable contract and compare volume, open interest, settlement prices, calendar spreads and roll dates.

Volume profile is not the live order book

Time-based bars answer “when did trading activity occur?” A volume profile asks “at which price bands did completed transactions accumulate within the selected sample?” High-volume nodes show areas of past acceptance; low-volume areas may indicate rapid traversal. Both describe history. They do not prove that the same orders remain available today.

The choice of starting and ending point materially changes the profile. Selecting a window after the full move is visible and then claiming that its high-volume node predicted the reversal creates look-ahead bias. A valid process fixes the window rule at the decision time, tracks how the distribution evolves and treats nodes as candidate areas for pause, acceleration or repricing—not automatic support and resistance.

There is no universal volume code across markets

Equities: venue, session and corporate actions come first

Equity volume may consolidate multiple venues or cover only one exchange. Auctions, regular trading and extended hours have different liquidity structures. Earnings, index rebalances, splits, offerings and block trades can reset the baseline. If a stock gaps higher on strong regular-session RVOL, confirmation includes holding the gap, closing in the new value area and checking adjusted data and free-float changes. If most of the print came from a closing auction or a one-off rebalance with no follow-through, an event explanation is stronger.

Futures: centralised reporting does not remove contract risk

Exchange futures volume is comparatively well defined, but the contract month must be named. Before treating an oil-futures breakout as participation confirmation, verify where volume and open interest migrated, whether the signal exists in the active contract, how the term structure moved and whether an inventory or settlement event distorted the bar. A jump visible only in a continuous chart around the roll is not a clean tradeable breakout.

Spot FX: tick activity is not global turnover

Spot foreign exchange is mainly an OTC market. Tick volume on one platform counts updates captured by that feed. It may correlate with market activity, but it is not worldwide traded value. After payroll data or a central-bank decision, a burst in ticks should be assessed alongside spread, slippage, completed bars and feed quality. Different platforms can show different extremes; one anomalous spike cannot represent the whole FX market.

Gold: separate spot, futures and regional venues

XAUUSD, the London wholesale market, COMEX futures and the Shanghai gold market have different hours, participants and reporting conventions. Strong COMEX volume demonstrates participation in that futures market, not a universal “inflow into gold”. A spot break supported by expanding futures activity can form one piece of cross-market confirmation. A brief break in only one series, especially when sessions do not overlap, deserves less weight.

Aggressor volume and cumulative indicators inherit data assumptions

Trade-classification rules assign executions to buyer or seller aggression and support tools such as volume delta or cumulative delta. They add information about which side appeared more urgent, but results depend on quote synchronisation, tick resolution, hidden orders, off-exchange trades and venue coverage. Buyer-initiated volume dominating while price fails to rise can indicate passive selling absorption; it can also reflect classification delay or incomplete data.

On-balance volume and similar cumulative indicators often assign an entire bar's volume a positive or negative sign according to the close. That is a compression rule, not a real fund-flow account. If a bar closes fractionally higher after intense two-way trade, every unit may be labelled positive. The appropriate question is whether the derived indicator adds stable out-of-sample information beyond price and properly normalised RVOL.

A six-step volume-analysis workflow

1. Define the market and feed: record instrument, venue, quote source, time zone, regular and extended sessions, and futures month.
2. Define the unit: shares, lots, contracts, turnover value, ticks, open interest and profile volume are not interchangeable.
3. Build a comparable baseline: use the median of N matched time slices and predefine treatment of holidays, events and unfinished bars.
4. Read price context first: trend, range, key zone, volatility and catalyst come before participation evidence.
5. Write confirmation and invalidation: for example, acceptance outside a zone, a quieter retest and renewed expansion; downgrade if price rapidly returns or volume was a one-off event.
6. Check executability: spreads, slippage, depth, auctions, price limits and roll costs may make a chart signal unavailable at the displayed price.

Use FastBull Charts to place price, volume, volatility and key zones on the same timeframe. If body, wick and close location are not yet familiar, revisit Part 1 on candlestick charts. When volume changes may be event-driven, verify the timing in the FastBull Economic Calendar.

Nine traps in volume backtests

1. Incomplete bars: generating an intrabar signal with partial volume but benchmarking it against completed bars.
2. Full-day look-ahead: using the final daily volume, closing auction or finished profile in a decision made earlier.
3. Missing seasonality: mixing opening and midday observations in one average and manufacturing anomalies.
4. Feed changes: mistaking altered venue coverage or cleaning methodology for a market regime shift.
5. Event and holiday contamination: allowing a few extreme sessions, half-days or closures to distort comparability.
6. Corporate actions: ignoring splits, consolidations, offerings, index additions and changes in free float.
7. Continuous-futures bias: combining expiries without modelling roll migration, spreads and settlement rules.
8. Parameter hunting: repeatedly changing lookbacks, RVOL thresholds and swing pairs until history looks persuasive.
9. Missing costs: omitting spread, slippage, market impact and unfilled orders when unusual volume coincides with volatility.

A useful report goes beyond the “win rate of a high-volume signal”. It should show median volatility-adjusted returns, maximum favourable and adverse excursion, liquidity buckets, event versus ordinary samples, holding period, invalidation rate and out-of-sample results after costs. If a small change of session or feed destroys the result, the strategy may rely on historical coincidence rather than a durable mechanism.

The 30-second conclusion: normalise first, interpret second

Volume's primary job is to describe activity, not to certify buying or selling. Identify whether the chart contains shares, futures contracts, traded value, ticks, open interest or volume by price. Establish a baseline using the same market, source and time slice. Only then place participation inside trend, location, volatility, liquidity and event context. High-volume breakouts, low-volume retests, divergences and rising open interest are pieces of evidence; they become testable only when paired with subsequent price acceptance and rules written before the result is known.

Risk warning: This article is for financial education only and is not investment or trading advice. Volume, open interest, divergences and historical backtests cannot guarantee future outcomes. Any decision must also account for objectives, risk tolerance, liquidity and transaction costs.