Anchored VWAP vs VWAP
Michael Whitman
·
Trading since 2017, full-time since 2020
·
Published 2026-09-08
·
Updated
·
Reviewed
Anchored volume-weighted average price and the session version run the identical calculation. The only difference is where the running sum begins: the session open in one case, an event you select in the other, and that choice is the whole technique.
Anchored volume-weighted average price and the ordinary session version are the same arithmetic. Price times volume, accumulated, divided by accumulated volume. The only thing that differs is where the accumulation starts, and on sixty sessions of SPY five-minute bars that one choice changed how often price crossed the line by a factor of between 5.7 and 7.5.
What each one is
Session VWAP starts at the 9:30 open and restarts each day. Everybody using the same data sees the same line, because the start is fixed by the session rather than by anybody’s judgment. VWAP covers the calculation.
The anchored version starts wherever you put it. A results release, a gap, a swing high or low, the day a policy changed. Anchored VWAP covers that variant and how to place an anchor.
Neither is smoothing price over a fixed window. Both are running averages weighted by how much traded at each price since the start point, which makes them a rough measure of what participants paid on average since then.
Where they differ
Where the sum begins. One restarts on a schedule; the other begins at an event and never restarts. Everything below follows from that.
How fast each line moves. Early in a session the session line rests on a handful of bars and moves quickly. An anchor set weeks ago has accumulated so much volume that one more bar barely shifts it. On SPY the session line moved several times further per bar, as the data section shows.
How often price crosses each. Because the session line restarts near the open price and tracks the day, closes cross it many times a session. An anchored line drifts away from price and stays there for days, so closes cross it rarely.
Where the judgment sits. The session version has none, which is a strength and a limitation. The anchored one is entirely a judgment about which event mattered.
How the running sum behaves
The formula is identical. Neither has a length setting; once the start is fixed, the line is determined. One choice that is often left unstated is the price used for each bar. Charting packages working from bars, rather than from every trade, commonly use the typical price, (high + low + close) ÷ 3, and that is what the figures on this page use.
Both weight by volume rather than by time, so a heavy bar moves them more than a quiet one, which is the property that separates either from an ordinary moving average.
A worked example of the session line. SPY’s first five-minute bar on 2 October 2026 had a high of $770.84, a low of $769.49 and a close of $770.70, so its typical price was $770.3433, and with 3,297,046 shares that is also the session VWAP after one bar. The second bar’s typical price was $770.7733 on 856,384 shares. The line after two bars is (770.3433 × 3,297,046 + 770.7733 × 856,384) ÷ 4,153,430 = $770.4320. The first bar carries almost four times the weight of the second, so the line moved less than a quarter of the way toward the new price.
An anchored line does the same sum from an earlier start. Anchored at the 10:30 bar on 13 August 2026, the line had 35 sessions of volume behind it by 2 October, and it ended that day at $766.08, while the session line ended at $769.90.
Which one to use: Anchored VWAP or VWAP?
Use the session version when your question is about today. It answers whether price is above or below what the day’s participants paid on average, and it restarts before each new day can be confused with the last.
Use the anchored version when a specific event is what matters. A results release, a policy change, the low a move began from. The session boundary is arbitrary in those cases; the event is not.
Use the anchored version for anything held longer than a day, because a line that restarts every morning cannot describe a position measured in weeks.
And when you cannot name why you put the anchor there, use the session version. An anchor without a stated event is a line placed where the chart looked right, which is fitting rather than analysis.
The original data
The sample. Yahoo Finance five-minute SPY bars for the 60 regular sessions from 10 July to 2 October 2026: 4,680 bars, 78 a session, after one 4:00 p.m. stamp outside the session was dropped. Sixty sessions in one stretch of one market is a small sample, so the figures describe this stretch and nothing wider. Both lines use each bar’s typical price and volume; neither is built from individual trades.
The anchors were picked in hindsight. One is the bar with the highest high of the 60 sessions, 10:30 on 13 August 2026 ($779.37, matched by the 10:35 bar; the first was used); the other is the bar with the lowest low, 3:55 p.m. on 29 July ($729.10). Each was measured from the next session onward, so every anchored value has at least one full day behind it.
| From the session after | Sessions | Bars | Close crossed session VWAP, per session | Close crossed anchored line, per session | Bars with the close between the lines |
|---|---|---|---|---|---|
| 13 Aug high | 35 | 2,730 | 7.26 | 0.97 | 42.78% |
| 29 Jul low | 46 | 3,588 | 7.09 | 1.24 | 47.97% |
A crossing is a close on the other side of the line from the previous close in the same session. The lines disagreed on a large share of bars. On 42.78% of bars after the high, the close was above one line and below the other, so a rule that reads “above VWAP” gave two answers on those bars. The median gap between the two lines was 0.48% of price.
Which side price sat on depended on the anchor. After the high, closes were above the anchored line on 31.58% of bars and above the session line on 45.64%. After the low, it was 74.92% and 48.19%. The anchored line inherited the direction of everything since its start; the session line did not.
How far each line moved per bar. The median five-minute change was 0.0019% of price for the session line after the high, against 0.0003% for the line anchored there. A bar of the day’s own volume is a large share of the session’s running total and a tiny share of 35 sessions’ worth. The session line’s median move was 6.3 times the anchored line’s after the high, and 10.5 times after the low.
The volume itself is approximate. Summed over the 78 regular-session bars, Yahoo’s five-minute volume came to a median 84.45% of Yahoo’s own daily volume for the same day (from 68.42% to 90.93% across the 60 sessions). The five-minute file does not say what the other sixth is. Trades before 9:30, after 4:00 p.m. or in the closing auction could account for it, and the 78 regular-session bars would not carry those, so the weights behind both lines may leave out part of the day’s trading. Every bar and both lines are in the SPY five-minute VWAP file, and the daily comparison is in the volume check file.
Where the anchored version goes wrong
The anchor gets moved until the line works. Nothing stops you, the chart improves visibly each time, and the finished line passes through points you already knew mattered. Both anchors above were placed at extremes that were only known once the 60 sessions were over, and a reader shown either line would see a tidy reference that nobody could have drawn on the day.
And a volume feed that misses part of the day affects both lines. Neither line shows how much volume it was built on, so it looks just as precise either way.
When it fails
The failure is the anchor chosen for its fit, and the resulting chart is genuinely convincing. You try a swing low; the line runs slightly wrong. You try the gap two weeks earlier; now it catches three reactions cleanly. That version gets kept. What has happened is a search over start dates, scored on how well the output matches history, and the line keeps fitting the bars it was chosen on and not the ones after.
The second failure is an anchor with no stated event. It is a fitted date.
A third is using the session version for a multi-day position. It restarts each morning.
A fourth is treating a cross of the session line as rare. On SPY’s five-minute bars it happened about seven times a session.
A fifth is taking a stop from either line. Neither contains one, and on these bars price closed through the session line several times a day.
And a sixth is trusting the weighting without checking the volume feed. On these bars it summed to about 84% of the reported daily total.
Related
Anchored VWAP covers the variant and how to place an anchor. VWAP covers the session calculation. And volume-weighted moving average is the neighboring idea with a fixed lookback instead of a start point.
Name the event before placing the anchor. Anchored to a bar picked because price later respected it, the line runs through a point already known to matter; anchored to an event before anyone knows what happens next, it is a genuine reference.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.