Anchored VWAP: What the Age of the Anchor Does to the Line

An anchored VWAP has one input a trader actually sets, and it is not a length or a smoothing factor. It is a bar. Everything the line does afterwards follows from how far back in the chart that bar sits, and the further back it is, the less any new bar can move the result.

That property is arithmetic rather than opinion, and none of the widely read explanations of the tool states it.

What the Line Averages, and Over Which Window

Two running totals sit behind every point on the line. One adds up price multiplied by volume, bar after bar. The other adds up volume alone. The value plotted at any bar is the first total divided by the second.

Both totals begin at the anchor and never restart. That is the whole of the difference from the session version, where both totals are thrown away at the open and rebuilt from nothing, and where the day is the window by construction.

So the anchored form is not a variant with an extra option. It is the same calculation with the window handed to the user, and a window has to start somewhere. Choosing the bar is choosing which trades count.

One consequence is worth having in mind before anything else. Because nothing is ever discarded, the window only grows. Every bar that passes is added to a total that already holds everything before it, and the totals themselves are what the next section is about.

The Newest Bar Is Worth Its Share of All Volume Since the Anchor

Work out what one new bar does to the plotted value. Before it arrives the line stands at some average. The new bar carries its own price and its own volume, and both totals grow by that bar contribution.

The result is that the line moves toward the new bar price by a weight, and the weight is the new bar volume divided by the total volume accumulated since the anchor.

Put ordinary numbers through it. If volume per bar is roughly even, a line anchored twenty bars ago moves toward a new bar by about one twentieth of the gap. Anchored five hundred bars ago, the same bar shifts it by about one five-hundredth.

Neither figure is a property of the market or of the instrument. Both are properties of how long ago the anchor was placed, and they hold on any timeframe and any symbol, because the denominator is a running sum that cannot shrink.

A heavy bar can override this for a moment, since the weight is a volume share rather than a bar count. But it takes a bar carrying a meaningful fraction of everything traded since the anchor to do it, and the older the anchor, the rarer such a bar becomes.

A Line Anchored Last Week and One Anchored Last Quarter Are Different Instruments

Two anchored lines on one chart are commonly read as two views of the same tool. The arithmetic above says otherwise. They are two weighted means over two different windows, and the older one is the stiffer of the two by a factor of however many bars separate the anchors.

That changes what a crossing means. When price crosses a line anchored a few bars back, both quantities are moving toward each other, and the crossing carries information about the average as well as about price.

When price crosses a line anchored a quarter ago, the average has barely moved at all. The event is a move in price against a nearly fixed level, and calling it a crossing dresses up something simpler than it sounds.

The same asymmetry applies to the two lines converging or separating. A gap between an old anchor and a recent one closes mostly because the recent one travelled, not because the two met in the middle.

None of that makes the older line worse. A nearly fixed level derived from every trade since an event is a perfectly reasonable thing to plot. It is simply not behaving like an average any more, and treating it as one is where the reading goes wrong. Whether a tool of this kind is describing the past or anticipating anything is the older question of whether an indicator leads price or follows it.

No Setting Makes It More Responsive

A trader who finds the line unresponsive will look for a setting. On a moving average there is one: shorten the period. Here there is nothing to shorten.

One vendor publishes the complete input list for its anchored study. It holds a start date and two numbers controlling how far the upper and lower bands sit from the line. The bands are drawn around the result and change nothing about the result itself.

So the input list is effectively one item long. Responsiveness is not exposed, not because the vendor withheld it, but because the calculation has no such term. The only way to change how the line behaves is to move the anchor, which changes the question rather than tuning the answer.

That is a sharper constraint than the explanations imply. Placing several anchors and reading them together is often described as flexibility, and it is really a workaround for a tool with a single knob, the same way reading one instrument over more than one window compensates for any single window being a choice.

The Anchored Version Has No Reset, and the Vendor Says So

The session form restarts. The anchored form does not, ever, until the user moves the anchor or removes it.

One vendor states this as a routing instruction inside its own study documentation: if the calculation should reset daily, weekly or monthly, the ordinary session study is the one to use instead. That single sentence settles what the two tools are for more directly than any comparison of their outputs.

The published explanations do mention the session reset, but they use it to describe what the ordinary version does before moving on. Framed the vendor way, the absence of a reset is the defining property rather than a point of contrast, and it is what makes the window grow without limit.

It also means an anchored line left on a chart quietly ages. Nothing on screen announces that its window is now four times longer than when it was placed, and the arithmetic above says that is precisely when its behaviour has changed most.

Read down the last row and the practical difference is clear enough. Everything the session form settles by convention, the anchored form leaves open, and the one thing it leaves open is the one thing that governs how the line behaves for as long as it stays on the chart.

Who Built It, and What the Explainers Never Name

The anchored tool has a documented origin. One charting vendor credits it in its own support pages to Paul Levine, a physicist and technical analyst, developed as part of a system of market analysis he worked on in the mid-1990s.

That name appears in none of the four explanations read for this page. One of them instead dates the plain volume-weighted average to a journal article of the late 1980s, without a citation that can be followed, and does not distinguish the origin of the plain form from the origin of the anchored one.

The distinction matters for a reader trying to check anything. The two tools were introduced decades apart by different people for different purposes, and a page that dates one while describing the other has answered a question nobody asked.

Only the vendor attribution appears here, and only because it is published by the vendor that ships the tool. It is not repeated as a fact about the history of technical analysis, because a support page is evidence about that vendor documentation and nothing more.

What the Volume Input Is on a Currency Chart

Everything above assumes the volume total is counting something transacted. On an exchange-traded instrument it is.

Spot currency has no central tape, so a retail terminal usually plots the number of price changes in the bar instead. The weighting still works arithmetically, but what it is weighting by has changed, and the page on what a tick volume field counts sets out that substitution in full.

The narrow consequence for this tool is the denominator. A weight expressed as a share of a total survives the substitution unharmed, because both the numerator and the denominator come from the same series, so every statement in this page about how the line stiffens with age still holds.

What does not survive is any reading of the line as a cost basis. Calling the value the average price paid requires the weights to be amounts transacted, and a chart drawn on a genuine traded-volume series is what that reading needs underneath it.

Which Anchor Belongs to Your Question

The choice is not between a good anchor and a bad one. It is between questions, and each question has an anchor that answers it.

A question about who is offside since a specific event, a release or a gap, is answered by an anchor on that event and by nothing else. The line will be responsive at first and will stiffen as the event recedes, which is appropriate, because the event recedes too.

A question about the level a long accumulation was built at is answered by an old anchor, and its stiffness is the point rather than a defect. Read it as a level, not as an average that is about to move.

A question about what has happened this week is not an anchored VWAP question at all. It is a session or rolling calculation, and reaching for an anchor there produces a line whose window keeps growing past the period being asked about.

Two habits follow from that. Write down what the anchor was placed on, because a line with no remembered reason becomes a level nobody can evaluate. And date it, since the age of the anchor is the single number that tells you how much of any crossing was the line and how much was price.

Neither habit is a technique. Both exist because the tool records the reason for its own window nowhere, and the trader who placed it is the only record there is.