What is the VWAP indicator?

VWAP is a cumulative average, not a rolling one, and almost everything surprising about it follows from that. For each interval of the session you take a typical price, conventionally the average of that interval's high, low and close, and multiply it by the volume traded in the interval. You keep a running total of those price-times-volume products from the open, and a separate running total of volume. VWAP at any moment is the first total divided by the second. Every new bar adds to both sums and nudges the line; nothing ever drops out, because there is no window to drop out of. The average simply grows heavier as the day accumulates volume.

That is the whole definition, and it is worth watching it work rather than taking it on trust. The figure below builds one session bar by bar. Each 30-minute interval contributes a dot at its typical price and a bar of volume beneath it, and the gold line is the actual cumulative VWAP of those bars, not a sketch of one. Watch the 11:45 interval in particular: price spikes twelve rupees on 4 lakh shares, the lightest bar of the day, and the line barely notices.

The arithmetic is easier to trust when you can see the columns. The table works the same session through, interval by interval, so you can follow both running totals to the close. Two rows are worth pausing on. The 11:45 row is the spike: it contributes the highest typical price of the whole day and the smallest volume of the whole day, and you can watch the running totals decline to care. The 15:15 row is the close: on its own it carries 52 lakh shares, nearly a fifth of everything traded, and it drags the line up more in one interval than the previous four did together.

Read the last column down and the character of the line becomes obvious. VWAP moves a long way early, when the divisor is small, and then settles into a slow drift as the totals get heavy. It never chases price, because it is not trying to: it is reporting the average rupee actually paid per share so far today. Everything else on this page is a consequence of the two design choices buried in that formula, so it is worth taking them one at a time.

Start with the word in the middle of the name. A moving average treats every bar as one vote. VWAP does not: it weights each price by the number of shares that actually changed hands there. A bar where a great deal traded counts for a great deal; a bar where almost nothing traded counts for almost nothing. That sounds like a small technical preference until you see what it does, because it changes what the number means. An unweighted average answers "where did price go today." A volume-weighted average answers "what did the average share actually cost today," and those are not remotely the same question.

The cleanest way to prove it is to hold price completely still and move only the volume. In the figure below, the two panels contain the identical twelve-bar price path. Every high, every low, every price is the same on both sides. The only thing that differs is which bars carried the size: on the left the heavy volume arrived on the dips, on the right it arrived on the rallies. The plain unweighted average of those twelve prices is therefore identical too, which is why the dashed line sits at exactly the same height in both panels. VWAP is not identical. It is seven and a half rupees apart.

A price only counts for as much as the volume that agreed with it. That single sentence is what the word weighted is doing in the name.

This is why VWAP is difficult to push around, and why it is treated as a fair-value reference rather than a line on a chart. To move it you have to trade real size, and trading real size is expensive. A thin spike, a stop run, a few large prints in an illiquid instrument at lunchtime: these move price, sometimes violently, and leave VWAP almost where it was, because the market as a whole did not agree to do business up there. Conversely, a level where 50 lakh shares changed hands is a level the market genuinely accepted, and VWAP will sit near it whatever the closing print says. The line is a record of consensus by size, and that is a far more robust thing than a record of where the last trade happened to land.

The second choice is the anchor. VWAP does not look back a fixed number of bars; it looks back to a fixed point, the session open, and cumulates forward from there. That has an immediate and underappreciated consequence: the divisor only ever grows. At 09:15 the sums are empty and the very first bar sets the whole line by itself. By 15:15 the sums are carrying the entire day, and a new bar has to argue with all of it. VWAP is not equally responsive through the session. It starts as an opinion and ends as an institution.

The figure below measures that directly on the same worked session. Take one identical bar, 10 lakh shares printing five rupees away from the line, and ask how far it actually moves VWAP depending on when it lands. At 09:15 it shifts the line by eighty-nine paise. At 15:15 the identical bar shifts it by eighteen. Nothing about the bar changed; only the size of the divisor it was fighting. This is the honest reason VWAP lags most in the afternoon, and it is not a defect to be tuned out. It is what anchoring means.

The other half of the anchor is the reset, which is the property most often misused. Because the cumulation runs from the open, at the next open the totals return to zero and a fresh line begins from that morning's first bar, wherever it happens to be. In the construction figure the old line ends at 1,857.80 and the new one starts at 1,847.33, ten rupees lower, because the market gapped down overnight and yesterday's accumulated volume has no bearing whatsoever on today's average cost. Carried across days, a standard VWAP is not merely less useful; it is meaningless. It is an intraday, session-anchored measure by design, and every property people admire in it, its resistance to spikes, its authority as a fair-value read, depends on the anchor being a real event rather than an arbitrary bar count.

People reach for VWAP and a moving average as though they were interchangeable smoothing lines, and the two properties above are exactly why they are not. A simple moving average inverts both choices. It weights every bar equally, so a 4 lakh bar and a 52 lakh bar cast the same vote. And it is anchored to nothing: it looks back a fixed number of bars from wherever you are standing, so the oldest bar drops off as each new one arrives and the window slides continuously, straight through the close, straight through the next open, indifferent to the fact that a new session began.

Set them on the same data and there is nothing left to argue about. Both panels below plot the identical 39 bars across three sessions. Above, VWAP appears as three separate lines, each anchored at its own open, each stopping dead at the divider. Below, the moving average is one continuous line that slides through both dividers as if they were not there, and the highlighted window shows it doing something VWAP structurally cannot: averaging four bars of yesterday with six bars of today, because a window measured in bars has no idea what a session is.

The practical consequence is that the two are not substitutes, and the only real error is to read one as though it were the other. A moving average will run happily on a daily or a weekly chart, where VWAP has nothing whatsoever to say, because out there no session exists to anchor to. VWAP will tell you what the average participant actually paid today, which no lookback average can, because a lookback average never asked how much traded. So the choice between them is not about taste, or about which reacts faster. It is about whether the question in front of you concerns a trend or a day's business. The table sets the differences out property by property.

Here is the fact that gives VWAP its authority, and the one most descriptions leave out. VWAP is not primarily a line traders watch; it is the benchmark large desks are measured against. When a fund needs to buy or sell a position too big to execute in one go, the quality of that execution is judged by comparing the average price achieved with the day's VWAP. A buy filled below VWAP, or a sell filled above it, shows the desk transacted better than the session average and, crucially, did not move the market against itself. That single comparison, the VWAP slippage, is a standard measure of execution quality across the buy side, and it is why VWAP is embedded in how institutions actually trade rather than in how they read charts.

To hit that benchmark a desk does not fire the parent order into the market at once, which would consume the resting liquidity on one side of the book and push price against itself. Instead an execution algorithm slices the parent order into many small child orders and releases them through the day on a schedule built to match the market's volume profile, so the trade blends into natural flow. The mechanism is worth stating plainly, because it is prettier than it first looks: if your slices are poured into the market in the same proportions the market itself trades, then your average cost converges on the volume-weighted average cost almost by construction. The algorithm is not forecasting anything. It is buying the day in the shape the day traded.

The line attracts flow because flow is measured against it. VWAP is not a magnet because of any property of price. It is a magnet because of a scorecard.

This is the point that turns VWAP from a chart curiosity into a market-structure fact. Because a large body of flow is deliberately referenced to VWAP, the line is partly self-referential. Aggregate enough benchmarked programs, all leaning their buying toward the line when price is below it and their selling toward it when price is above, and you get something that behaves like a soft restoring force: deviations get leaned on, which is why price is anchored to VWAP intraday and why the line reads as fair value. Note carefully what this does not say. It does not say price must return to VWAP; it says there is a population of orders with a reason to lean that way, and populations of orders can be overwhelmed. Exactly how that feedback produces the reversion tendency, and precisely when it breaks on a trend day, is the mechanism the VWAP and mean reversion guide takes apart in full.

The Indian equity cash session runs from 09:15 to 15:30, and there is a detail inside it that settles any argument about whether VWAP is a serious institutional object. The official closing price of a share has not been the last trade of the day. It has been a volume-weighted average price: the exchange takes every trade in the final thirty minutes, between 15:00 and 15:30, multiplies each price by its quantity, and divides by the total quantity traded in that window. The number that goes into your holdings statement, into index computation, into derivatives settlement and into mutual fund NAVs is a VWAP.

The reasoning is the volume-weighting property from earlier in this page, applied where it matters most. A closing price set by the last print would be trivially cheap to move: one small trade at 15:29:59 could stamp a number onto every portfolio in the country. A volume-weighted average over thirty minutes cannot be shifted that way, because to move it you would have to trade genuine size against everyone else trading in the same window. The same robustness that makes VWAP resistant to a thin spike on a chart is what made it fit to carry the official close.

That mechanism is now changing, and the change is worth knowing because it is imminent and because of what it concedes. Under a SEBI circular of January 2026, from 3 August 2026 stocks with active derivative contracts move to a closing auction session: regular trading in those stocks ends at 15:15, a 20-minute call auction runs to 15:35, and the close becomes that auction's equilibrium price, the level where the most volume finds a match. Stocks without derivative contracts stay on the 30-minute VWAP for now. The concession is in the detail: the auction's own reference price is taken as the VWAP of trades between 15:00 and 15:15. Even the mechanism designed to replace the volume-weighted average close needs a volume-weighted average to anchor itself to.

Read as intraday fair value, VWAP gives a quick and honest sense of the day's balance. Price trading above VWAP says that, on balance, buyers have paid up against the session's average cost; price below VWAP says sellers have pushed it under. Many platforms extend this with standard-deviation bands, lines plotted a set number of standard deviations of price above and below the line, forming a dispersion envelope: a touch of an outer band flags an unusually large deviation from the day's average. All of that is context. It describes the session so far. None of it is a forecast of the next move, and the four limits below are where the description runs out.

They are worth naming precisely, because each one is a direct consequence of something earlier on this page rather than a flaw someone could patch. VWAP is backward-looking because an average of what has already traded cannot be anything else. It resets daily because the anchor is the open. It lags late in the session because the divisor only grows, which the third figure measured at roughly five times less influence by the close. And it needs real volume to mean anything, because volume is the entire weighting scheme: on a thin instrument a handful of large prints can drag the line somewhere the market never really traded, and the number inherits an authority it has not earned.

VWAP belongs to the context layer of intraday work, and it is unusually good there for a reason worth restating: it is one of the few lines on a chart that is not an opinion about price. It is a measurement of what was paid, weighted by how much was paid at each level, from a fixed and meaningful anchor. That is why it survives contact with institutions, why an exchange was willing to build a closing price out of it, and why a desk will accept it as the number their year is judged on. It earns its authority from arithmetic and from the flow that references it, not from a claim about the future.

A reference tells you where you are. It was never going to tell you where to go, and no amount of standard-deviation bands will change that.

Which is exactly where its usefulness ends and the actual work begins. VWAP will not tell you whether a level deserves a trade, where the idea is invalidated, or how much of the account to put behind it. Those judgements come from reading the session's structure, the volume behind the moves, and the levels sitting around the line, and that upstream work is what the method we teach is built around. Learn the construction well enough to know what the line can and cannot see, and VWAP becomes a genuinely load-bearing part of an intraday read. Treat it as a signal generator and you have taken a measurement and asked it for a prediction it was never built to make.

• VWAP construction and the daily reset. The formula, cumulative sum of typical price times volume over cumulative volume, the typical price as the average of high, low and close, and the reset at each session open that makes VWAP an intraday measure. stockcharts.com
• VWAP as an execution benchmark. The use of VWAP as the standard against which large-order execution quality is judged, with algorithms slicing a parent order into child orders through the day and VWAP slippage measuring the result. en.wikipedia.org
• Anchored VWAP. The identical calculation begun from a bar the trader picks rather than from the session open, and then run forward across as many sessions as the question needs. stockcharts.com
• Backward-looking nature and late-session lag. VWAP reflects past price and volume, and the closer to the close, the more lag the line carries as accumulated volume makes each new bar less able to move it. tradingview.com
• The Indian closing price and the closing auction session. The Securities and Exchange Board of India circular of January 2026 introduces a closing auction session in the equity cash segment from 3 August 2026 for stocks with active derivative contracts, replacing for those stocks the existing close based on the volume-weighted average price of the last thirty minutes, with the auction's reference price itself taken from the volume-weighted average of trades between 15:00 and 15:15. sebi.gov.in