What do volume-based indicators add that price alone can't?
Price tells you what was agreed. Volume tells you how many shares were involved in the agreeing. Volume indicators are the only family in this course that reads a series the price oscillators never see — which makes them the one genuine addition to a chart already full of momentum lines.
OBV: a running tally with a signature flaw
On-Balance Volume, popularised by Joe Granville in the 1960s, is a cumulative total:
- Close higher than the previous close → add the whole session's volume
- Close lower → subtract the whole session's volume
- Close unchanged → leave it alone
Worked example (illustrative volumes, starting the tally at zero):
- Day 1, close 100 — baseline, OBV = 0
- Day 2, close 102, volume 1.0M → OBV = +1.0M
- Day 3, close 101, volume 1.2M → OBV = −0.2M
- Day 4, close 104, volume 2.0M → OBV = +1.8M
What traders read is the slope. Rising OBV while price goes sideways says up-sessions have been carrying more volume than down-sessions over that stretch. Those are two pictures, because they are two series. Here is the price:
And here is the tally over the identical sessions:
Now the flaws, which are rarely taught. A close +0.01% adds the entire day's volume, exactly like a +5% close — only the sign matters, so magnitude is discarded. The absolute level is meaningless, because the starting point is arbitrary; two charts begun on different dates show different OBV levels for the same stock. And it is path-dependent, so it can only ever be read as a shape, never as a value.
Volume-weighted ideas: VWAP
VWAP = Σ(price × volume) ÷ Σ(volume) across a session, usually using each bar's typical price, (High + Low + Close) ÷ 3.
Worked example. Three fills: 1,000 shares at 10.00, 3,000 at 10.10, 6,000 at 10.30.
- Value traded = 10,000 + 30,300 + 61,800 = 102,100
- Volume = 10,000 shares
- VWAP = 102,100 ÷ 10,000 = 10.21
The plain average of the three prices is 10.13. Volume weighting drags the number toward where the shares actually changed hands — which is the whole idea. VWAP is where the average share traded, not where the average price was.
That is why institutions care: a desk working a large order is frequently measured against the day's VWAP, which connects directly to the "big feet" idea from Foundations. Standard VWAP resets each session, so it is an intraday benchmark; anchored variants start the sum from a chosen event instead.
The caveats on the data itself
Volume is messier than price. Trading is fragmented across venues, a share of activity prints off-exchange, ETF creation and redemption flows distort single days, and reporting conventions differ by market. And volume confirms nothing on its own — a record-volume session simply means many shares changed hands, which is equally true at tops and at bottoms.
In the data
Volume is not always counted in the same unit. For a stock it is shares. For Bitcoin against the dollar, the daily volume in this data is already a dollar amount. Read the scale on the chart below: over the year to 28 September 2026 the bars ran from about 9 billion to 153 billion a day, with a median near 38 billion. As a count of coins that would be absurd, since fewer than 21 million bitcoin can ever exist; as dollars it is plausible for a coin that traded between about $80,900 and $87,400 in the week of 21 September 2026.
So the usual step from volume to value traded, shares times price, counts the price twice on a crypto pair and produces a number in the quadrillions. Before you compare participation across instruments, check what the volume is counted in.
Try it now
- Describe the pair of schematics above in one sentence that mentions both. Then note what you cannot say from them: the tally's level is whatever it was started at, so only the slope carries anything.
- Now look at real participation. Find the highest bar in the year of volume below, note its date, and find out what was happening that day — earnings, an index rebalance, a news event. Volume spikes almost always have a boring explanation, and the date is all you need to go looking. In the twelve months to September 2026 that bar is 26 June 2026: the tables under the chart are the headlines the feed carried about Apple that day, Apple's reporting dates around it, and its ex-dividend dates. If none of them explains it, name the scheduled event the date itself suggests: it was the last Friday of June.
- Compute a small VWAP by hand from three bars using price × volume — take the last three sessions of the table below, with each close standing in for the day's price — then compare it with the simple average of those three prices and note which direction the weighting pulled.