VWAP Strategies: 4 Setups, the Band Math, and 208 Band Fades Counted Two Ways

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-05 · Updated 2026-11-02 · Reviewed 2026-11-01

Covered on this page: TradingView.

VWAP strategies are four different trades built on one line: a long-or-short filter, a fade of the bands, a buy on the first return, and an anchored version that starts at an event. The band fade shows why the count matters: across four US index funds this autumn, most fades came back to the line, but most hit a tight stop first.

How it works - four different trades

One: the filter. Only take longs above the line, only shorts below it.

Two: the band fade. Sell the upper standard-deviation band, buy the lower, target the line.

Three: the first return. Wait for price to leave the line and buy the first time it comes back.

Four: the anchor. Start the calculation at an event rather than at the session open.

They share a line and nothing else. One is a condition, one is mean reversion, one is a pullback entry, and one is a different measurement altogether. The line itself is the volume weighted average price: price times volume, summed from the open, divided by the volume summed from the open.

The filter and its hidden premise

Picture a session that drifts upward while swinging back and forth around its own VWAP. Take the same long trade, with the same stop and target, from bars that closed above the line and from bars that closed below it.

On that kind of day the standard rule can run backwards. Buying below the session average is buying dips in an uptrend, and buying above it is buying strength that has already run.

In a session that trends hard, the reasoning goes the other way: price that leaves the line keeps going, and the entries above it are the ones the rule was written for.

So “long only above VWAP” is a condition whose usefulness depends on whether the day is trending or reverting — which is a market structure question, not a VWAP one. This page does not test which kind of day SPY had more often.

The premise is the part to check on your own instrument, with costs included, before the rule is trusted.

The first return

The argument for the first return is about orders. The first time price comes back to the line, traders who missed the move and traders who want to add may both be waiting there; by the fourth return, many of them have already acted.

It is the same argument made for any level. The support and resistance page makes it from a different direction. It is a reason, not a measured edge, and nothing on this page counts first touches against later ones.

Why the line stiffens

The denominator is the session’s running volume. Each new bar pulls the line toward its own price by exactly that bar’s share of the volume so far. Early in the session the running total is small, so one bar moves the line a long way; by midday it is large, and the same bar barely registers.

On SPY, the shift was steep. Across 38 sessions from 4 Aug to 23 Oct 2026, the bar at 9:35 a.m. carried a median 21.82% of the session’s volume to that point, and the line moved 21.82% of the way to that bar’s price. The bar at 3:00 p.m. carried a median 1.52%.

Then it loosens again. Volume picks up into the close, and the 3:55 p.m. bar, the last of the day, carried a median 8.29%. A late cross of the line is therefore not always a cross of a fixed level: in the last five minutes the line moves toward price noticeably again.

Two consequences. A midday cross is mostly price moving, because the line barely moves to meet it. And a flat midday line is not a stronger level. It is a less mobile one, which is a different thing being read as significance.

A worked example

SPY on 23 Oct 2026. The 9:35 a.m. bar traded 507,627 shares, with 2,630,709 traded so far that session, so it carried 19.30% of the weight. Its typical price, (high + low + close) ÷ 3, sat $0.68 under the line, and the line fell $0.13 toward it.

At 3:00 p.m. the same session had traded 21,783,646 shares. The new bar’s 239,041 shares were 1.10% of that. Its typical price sat $0.46 under the line, and the line fell about half a cent.

The four do not mix. The filter says do not sell strength; the band fade says sell it.

The original data

The sample: SPY five-minute bars from Yahoo Finance for the 60 sessions from 3 Aug to 26 Oct 2026. Twenty-two sessions were left out because they contain bars with prices but no volume (30 bars in all), which would break the running total. That leaves 38 full sessions of 78 bars, and 2,926 measured bars after each session’s first. Session VWAP uses the typical price, (high + low + close) ÷ 3.

The first block carried more than twelve times the weight of the 1 p.m. block (13.64% against 1.08%). The share is also exactly the fraction of the gap between the line and the new bar’s price that the line closes, which the script checks on every bar. Every bar, its volume, the running total and the line are in the bar-by-bar file.

Checked against Nasdaq’s daily SPY history: on 37 of the 38 sessions, the five-minute high and low were within 5 cents of Nasdaq’s. The five-minute volume added up to a median 84.62% of Nasdaq’s daily volume (76.06% to 90.45%), so part of each day’s volume is missing from the bars, and the shares above describe the bars Yahoo reports. The last five-minute close was within 5 cents of Nasdaq’s close on only 18 of the 38 days (largest gap $0.11), because the official close is set in the closing auction at 4 p.m.

Thirty-eight sessions from one autumn is a small sample of one fund. The shape follows from the arithmetic; the exact percentages belong to these weeks.

In the corpus of 24,971 trading videos, 62 have “VWAP strategy”, “VWAP strategies” or “VWAP trading” in the title, from 49 channels, at a median of 22,474 views; 40 of them (64.52%) stay under 50,000.

Second test: the band fade, counted both ways. We took Yahoo Finance 5-minute bars for SPY, QQQ, IWM and DIA, the last 60 sessions to 30 October 2026, and kept only sessions with all 78 bars and volume on every bar: 39 for SPY, 21 for QQQ, 53 for IWM and 57 for DIA. In each one, the signal was the first bar from 10:00 a.m. on to close beyond a 2 SD band, at most once per side.

The first count asks whether any later bar that day reached VWAP. The second puts a stop one standard deviation past the entry close and asks which came first, the stop or the line.

One signal reached both on the same bar, so its order is unknown.

Same signals, two pictures. 157 of the 208 came back to VWAP before the close, 75.48%. As trades, 138 (66.35%) hit the stop first, and 94 of those reached the line later, after the stop had already gone. 62 (29.81%) reached VWAP first.

By fund, 43 of 55 SPY signals came back, 16 of 25 for QQQ, 43 of 60 for IWM and 55 of 68 for DIA.

What this does not say. At entry the line sat at least two deviations away and the stop one, so the stop-first share is not a profit figure on its own, and no costs are included.

A wider stop changes the split. Yahoo is the only source for these bars; 170 sessions from one autumn is a small sample. Every signal is in the signal file.

Anchoring

The reset is a choice. Start the sum at a session open and you get the session VWAP; start it at a gap, an earnings bar, or a swing low and you get the average price paid since that event.

The catch is that you chose the anchor. A different anchor gives a different line and the same bar can read above one and below another, so an anchored VWAP is only as defensible as the reason for the anchor, and the reason has to exist before the line is drawn.

That split is common in practice: on SPY five-minute bars after the 13 August 2026 high, the close sat above one line and below the other on 42.78% of bars, as the anchored VWAP vs VWAP comparison measures.

How the bands are built, and fading them

The bands are a standard deviation, weighted by volume, counted from the same open as the line. Keep three running sums from the first bar: volume, typical price times volume, and typical price squared times volume. The line is the second sum divided by the first. The band width is the square root of (third sum ÷ first sum) minus the line squared.

Two widths above the line is the upper 2 SD band, two below is the lower one. That is the definition this page’s test uses; some charting tools offer other band modes, which the VWAP bands page covers.

The fade sells a close above the upper band, or buys one below the lower, and aims at the line. The trouble is in how the result gets counted.

A real one: SPY on 30 October 2026. The 10:05 a.m. bar closed at $759.59, under the lower band at $760.04, with VWAP at $760.42 and one standard deviation worth $0.19. A long there with a stop one deviation lower put the stop at $759.40, 19 cents away, and the target 83 cents away.

The next bar traded down to $758.57 and took the stop. At 10:35 price was back at the line, so “it came back to VWAP” is true. The session still closed at $755.71, $3.88 under the entry.

Counted as a return to the line, it goes in the yes column; counted as a trade, it is a loss. Only the second is something an account feels.

When it fails

The market ranges

Every crossing is a signal to a filter that only knows which side price is on. In a range the line is crossed again and again, so a filter flips each time and a strategy that respects it does nothing but reverse.

It was the wrong session

A VWAP anchored to the wrong open is a line with no meaning at all. Instruments that trade around the clock have no natural session boundary, which is why the trading sessions page matters more here than it looks.

You read the flat line as strength

The stiffening is the trap. A line that has stopped moving at midday looks like a level being defended. On SPY a new bar there carried about 1% of the weight, so the flatness is mostly arithmetic.

You forgot the close loosens it

The last bar is heavy. With a median 8.29% of the weight on the 3:55 p.m. bar, the line can move toward a late push in price, and a cross that looked clean at 3:50 can be gone by the final print.

Questions people ask

What is the best VWAP strategy?

There is no single best one, because the four are different trades. The filter is a condition, the band fade is mean reversion with a stop problem, the first return is a pullback entry, and the anchored line measures something else. The useful question is which one matches the kind of day you are trading.

Why does VWAP move less later in the day?

Because each bar moves the line by its share of the session’s volume so far. On SPY in 2026, that share was a median 13.64% for bars before 10 a.m. and 1.08% for bars between 1 and 2 p.m.

How are VWAP bands calculated?

From the open, as a volume-weighted standard deviation of the typical price around the line. On SPY at 10:05 a.m. on 30 October 2026 one deviation was $0.19, so the 2 SD bands sat 38 cents either side of VWAP.

What happens when price closes outside a VWAP band?

On SPY, QQQ, IWM and DIA from August to October 2026, 157 of 208 first closes beyond a 2 SD band came back to the line the same day, but a stop one deviation away was hit first on 138 of them.

Does VWAP reset every day?

The session version does, at the open. An anchored VWAP starts wherever you choose and keeps running until you move the anchor.

Related

VWAP is the line itself — the calculation, and why volume weighting changes it.

Mean reversion is the premise the band fade rests on, and what it costs when the premise is wrong.

And day trading is where all four of these live, because every one of them resets tomorrow.

This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.