Most indicators are descriptions of price. VWAP is different in kind: it is a number large participants are measured against, and that changes behaviour around it. Understanding why institutions care about it explains almost everything about how price acts near it — and saves you from the two standard ways retail traders misuse it.
What it is, mechanically
Volume-weighted average price is exactly what the name says: every trade of the session, weighted by its size. It is the session's average done business. Nothing about it is predictive by construction — it is a running record of where the session's volume actually changed hands.
Its importance is institutional, not technical. Execution desks are routinely benchmarked against it: an order worked through the day is judged by whether it filled better or worse than the session's VWAP, and whole families of execution algorithms exist to track it. That means real money is systematically working orders relative to this line all session long.
Why price behaves around it
A buyer benchmarked to VWAP is structurally patient above it and structurally active at or below it — filling below the benchmark is what a good execution looks like. Multiply that by every desk working orders in the session and you get the familiar behaviour: on rotational days, price stretched away from VWAP tends to get faded back toward it, because the stretch is precisely where benchmarked flow has the least reason to chase and the most reason to wait.
The standard deviation bands most platforms draw around VWAP quantify that stretch. One sigma is ordinary variation. Two sigma is a statistically unusual distance from the session's done business — and what happens there is one of the cleanest regime reads available:
- Balanced day: a 2σ stretch attracts responsive sellers (or buyers), volume confirms the fade, and price rotates back toward the average. The stretch was an error.
- Trend day: the same stretch finds initiative volume instead, the bands themselves start sliding, and reversion never comes. The stretch was information.
The band does not tell you which one you are in. The response at the band does. That is the honest use of the tool: not "price is at 2σ, therefore fade," but "price is at 2σ — now watch what participation does."
The two standard misuses
Treating VWAP as support. It is not a wall; it is an average. Price crosses it constantly, and on trend days it will be crossed once and never revisited. What is true is subtler: it is a location where benchmarked flow has reason to act, which makes the manner of a touch informative — a first touch after a long stretch is a different event from the ninth chop through it at lunch.
Carrying it across sessions. VWAP is a session construct because the benchmark resets with the execution day. An "anchored VWAP" from an arbitrary swing low is a legitimate study, but it is your anchor, not the market's — no desk is being measured against it. The session VWAP is the one with real flow behind it.
The practical rule
Read VWAP as the line where the session's argument is currently settled, and the sigma bands as how far the argument has strayed from settlement. Distance from VWAP tells you tension. The response at the extremes tells you regime. Neither one, alone, tells you direction — and any tool that claims otherwise is dressing an average up as a prophecy.