WhitmanTrading

VWAP Bands: The First Bars Mean Nothing

VWAP bands are standard-deviation bands plotted either side of the volume-weighted average price, using the deviation of price about that average across the same session. Because the deviation accumulates from the open, the bands are widest and least reliable in the first bars of a session.

How it works

A candlestick chart of the site's shared price history. The headline on the chart reads: Bands drawn around the volume-weighted average.
Bands drawn around the volume-weighted average. Illustrative chart - not real market data.

The centre line is the volume-weighted average price, and the bands sit either side of it. They sit at multiples of the standard deviation of price about that average, over the same session.

A gently rising stretch of the long price series. The headline on the chart reads: The centre is VWAP; the bands are its own deviation.
The centre is VWAP; the bands are its own deviation. Illustrative chart - not real market data.

The detail that matters is rarely stated: the deviation is measured about the weighted average, not a simple mean. And it accumulates: every bar since the open stays in the sum, so width depends on how much of the session has elapsed.

A calmly advancing stretch of the long price series. The headline on the chart reads: And it resets every session, which is the whole point.
And it resets every session, which is the whole point. Illustrative chart - not real market data.

Which is why the reset is the whole point. At the open the sums are emptied and the calculation begins again, so the first bar’s bands share no arithmetic with yesterday’s.

A choppy, directionless stretch of the long price series. The headline on the chart reads: So the bands are meaningless in the first few bars.
So the bands are meaningless in the first few bars. Illustrative chart - not real market data.

So the bands are meaningless in the first few bars. A deviation computed from a handful of observations is not a measurement, and the widest, most inviting bands of the day appear exactly when the sample is smallest.

How the edges are used

As the session fills out, the bands narrow and settle. That is the reverse of how most people trade them: the loudest stretch comes first, the trustworthy reading only once the day is half over.

A flat, quiet stretch of the long price series. The headline on the chart reads: The outer band is where a fade is usually taken.
The outer band is where a fade is usually taken. Illustrative chart - not real market data.

The outer band is where a fade is usually taken. On a rotational day price stretches to the edge, fails to hold, and turns back toward the centre — a mean reversion trade with a reference.

A strongly rising stretch of the long price series. The headline on the chart reads: And in a trend price rides it instead of returning.
And in a trend price rides it instead of returning. Illustrative chart - not real market data.

And in a trend price rides it instead of returning. Bar after bar closing along the outer edge is evidence of a trend day, not an extreme. The same touch means opposite things in the two states.

A declining stretch of the long price series. The headline on the chart reads: An anchored version starts from an event, not a clock.
An anchored version starts from an event, not a clock. Illustrative chart - not real market data.

An anchored VWAP version starts from an event, not a clock. The session open is an arbitrary boundary; anchoring to a gap or a news bar builds them around something that actually happened.

In practice

A candlestick chart with a volume histogram beneath it, with the volume histogram emphasised. The headline on the chart reads: It is the one overlay that actually uses participation.
It is the one overlay that actually uses participation. Illustrative chart - not real market data.

It is the one overlay that actually uses participation. Almost every other line here comes from price alone; this one has volume in the arithmetic, which is the difference from a band round a moving average.

A long-horizon candlestick view of the same price series. The headline on the chart reads: On a daily chart a session VWAP does not exist.
On a daily chart a session VWAP does not exist. Illustrative chart - not real market data.

On a daily chart a session VWAP does not exist. These belong to intraday trading; a platform drawing them on a daily series is anchoring arbitrarily, or accumulating for weeks.

A candlestick series containing several opening gaps, with the largest opening gap marked. The headline on the chart reads: And a gap open builds the whole thing somewhere new.
And a gap open builds the whole thing somewhere new. Illustrative chart - not real market data.

And a gap open builds the whole thing somewhere new. After an opening gap the calculation restarts nowhere near yesterday’s close, so yesterday’s bands carry nothing.

A declining stretch of the long price series, with the entry price and the level at which a stop would trigger drawn as horizontal lines. The headline on the chart reads: The far band is the only level it defines.
The far band is the only level it defines. Illustrative chart - not real market data.

The far band is the only level it defines. A fade at the upper edge puts the stop loss beyond it; a tighter one discards the structure that justified the trade.

A candlestick chart of the site's shared price history, annotated with the round-trip cost. The headline on the chart reads: Every fade costs 2% of a bar.
Every fade costs 2% of a bar. Illustrative chart - not real market data.

Every fade costs 2% of a bar. Round-trip cost on this site’s shared 576-bar history is 0.0098 price units — 2% of a median bar’s range, and 45% of the smallest bar.

A 72-bar candlestick section of the shared price history. The headline on the chart reads: Institutions are measured against it, which is why it works.
Institutions are measured against it, which is why it works. Illustrative chart - not real market data.

Institutions are measured against it, which is why it works. Execution desks are benchmarked to the volume-weighted average, so real order flow references the level, market makers included. That is more mechanism than most overlays have, and no more.

Working a session honestly

Ignore them for the first stretch. Not as a preference but as a rule: until the session has enough bars in it the band is a drawing, not a statistic, and nothing at its edge is worth trading.

Wait for the width to stop moving. It settles as the sum fills out, and once successive bars stop shifting it, the deviation describes the day rather than its opening minutes.

Then require something other than the touch. A rejection, a failure to extend, a level from elsewhere — the band says where price is unusual for this session, which is a location, not a reason.

And decide which state you are in first. The same edge is a fade on a rotational day and a hold on a trending one, so VWAP strategies skipping that judgement fail worst in the strongest markets.

What VWAP bands are not

When it fails

In a range every touch looks like a signal

A sideways, range-bound candlestick series. The headline on the chart reads: In a range every touch looks like a signal.
In a range every touch looks like a signal. Illustrative chart - not real market data.

A trading range delivers touch after touch, and most of them work. Then the range ends, the last one does not, and the losing trade is larger than the several that paid.

The opening stretch is the trap

The bands are widest when the sample is smallest, which is precisely when a fade looks most obvious and is worth least.

Band width is not comparable between days

The fourteen-bar average true range on the shared history runs from 0.2823 at the tenth percentile to 0.7954 at the ninetieth, a ratio of 2.82; bar ranges spread wider still, 0.17 to 1.101, a ratio of 6.5. A width that means stretched on one day is ordinary on the next.

The costs compound

Round-trip cost exceeds a tenth of a bar’s range on 15 of the 576 bars measured. Fading every touch pays it repeatedly, and does so on the days when ranges are tightest.

A trend day punishes every fade

The minority state is where this produces its worst losses, because a trending market holds price outside the band while each fade is stopped out at the far edge.

The original data

A 72-bar window of the shared price history, cut short at the decision bar. The headline on the chart reads: Price is at the upper band, trending. Fade it?
Price is at the upper band, trending. Fade it? Illustrative chart - not real market data.

478 videos in the corpus have VWAP in the title, across 204 channels, at a median of 4,113 views and a maximum of 4,351,269 — one of the most covered indicators on this site. The scan is research/broker-coverage.json, over the 31,760 videos in research/search-study-corpus.jsonl. The bands, which generate the entries people actually take, are not separately covered at all.

research/series-measurements.json, built by site/measure_series.py, gives the honest odds. The ten-bar efficiency ratio on the shared 576-bar history has a median of 0.34, with 30% of bars above 0.5 — three bars in ten trend, so the fade sits with the majority state and loses worst in the minority. Check how much of the session has elapsed before treating a band as a level, and never fade one in the opening stretch.

VWAP is the centre line these are drawn around, and the place to start if the weighting is unfamiliar.

Anchored VWAP removes the clock from the calculation, which fixes the arbitrary open that makes early bands unreliable.

Volume analysis is the wider subject underneath, since participation is what separates this from every price-only overlay.

What I actually do

The mistake I kept making was fading the upper band early in the session, because that is exactly when it looks furthest away and most obviously stretched. What I was actually trading was a band drawn from barely any bars, which will sit almost anywhere. Now I leave it alone until the width stops moving, and most of the trades I used to take simply stop appearing.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.