Bands in VWAP: how they are calculated and what a backtest shows
Bands in VWAP are lines a chosen number of standard deviations above and below the volume-weighted average price, reset at the session open or at an anchor event. They measure how far price has stretched from where most volume traded. Whether a touch means a reversal or a breakout is a rule to test after costs, not a property of the band.
By Soren Huang · Published · updated · Editorial policy
Contents
- What are bands in VWAP?
- How are VWAP standard deviation bands calculated?
- VWAP bands vs Bollinger Bands: what differs?
- Why does the session reset matter, and what is anchored VWAP?
- Which standard deviation should you use: 1, 2 or 3?
- Band touch as mean reversion or band break as trend: which rule to test?
- How much do costs and slippage take from intraday VWAP band trades?
- Frequently asked questions
- References
What are bands in VWAP?
Bands in VWAP are lines drawn a set number of standard deviations above and below the volume-weighted average price. They show how far price has moved from the level where most volume traded since the session or anchor began. The bands measure distance. Whether a touch means a reversal or a breakout is a rule you must test.
VWAP is the volume-weighted average price (VWAP): the average price of a session where every bar counts in proportion to the volume it traded. The VWAP bands indicator page covers the plotting details; this article spends its words on how the bands are built, where they differ from other bands, and what a test of a band rule looks like once commission and slippage are charged. A standard deviation is a single number that says how widely the prices of the session have spread around their average. One standard deviation is a normal spread; three is unusual.
Many intraday traders treat the bands as support and resistance. That is a hypothesis, and like every other level drawn from an indicator, it should be written as an exact rule before anyone trades it. The claim this article defends is simple: a VWAP band is a measurement, not a signal, and the same band supports two opposite rules. Only a test with costs tells you which one, if either, survives.
How are VWAP standard deviation bands calculated?
VWAP is the running sum of price times volume divided by the running sum of volume, and the bands add or subtract a volume-weighted standard deviation of price around that average, both reset at the chosen start point. Think of a scale: each bar puts a weight on the scale equal to its volume, at a position equal to its price. VWAP is the point where the scale balances. A bar with heavy volume pulls the balance point toward its price; a thin bar barely moves it.
The price in that running sum is usually the typical price of the bar, the average of high, low and close, and the volume is the contracts traded in that bar. Both sums grow bar by bar and never shrink until the reset. The standard deviation is built from the same running sums: each bar's distance from the current VWAP counts in proportion to that bar's volume, and the result is the volume-weighted spread of the session's prices around VWAP. The upper band at k standard deviations sits k times that spread above VWAP; the lower band sits the same distance below. Platforms differ in small ways, for example whether they use the typical price or the close, so check the setting on the platform you trade and match it in the backtest.
Hypothetical example
Three bars after the open. Bar 1: price 100, volume 10, price × volume = 1,000. Bar 2: price 102, volume 30, price × volume = 3,060. Bar 3: price 101, volume 20, price × volume = 2,020. The sums are 6,080 and 60, so VWAP = 6,080 ÷ 60 = 101.33. The volume-weighted standard deviation works out to about 0.75 points. The 1 SD bands sit near 100.58 and 102.08; the 2 SD bands near 99.83 and 102.83. Bar 2 pulled VWAP above the simple average of 101 because it carried half the volume.
Two consequences follow from the cumulative sums. Early in the session, a few bars dominate, so VWAP and the bands jump around. Late in the session, the sums are large, and a single bar can barely move them. The bands are therefore wide and unstable in the first hour and narrow and steady in the last hour, for reasons that have nothing to do with price behaviour.
VWAP bands vs Bollinger Bands: what differs?
Bollinger Bands center on a rolling moving average over a fixed number of bars and ignore volume, while VWAP bands center on a volume-weighted average that grows from a fixed start, so the two react to the same move at different speeds and for different reasons. The Bollinger Bands indicator uses the last N bars, typically 20, and drops the oldest bar every time a new one arrives. VWAP bands never drop a bar until the reset.
| Property | Bollinger Bands | VWAP bands |
|---|---|---|
| Center line | Simple moving average of the close | Volume-weighted average price |
| Lookback | Fixed number of bars (rolling) | Everything since the session open or anchor (cumulative) |
| Volume used | None | Every bar weighted by its volume |
| Width late in the session | Tracks the last N bars; can widen again at any time | Tends to narrow and settle as volume accumulates |
| Same value on every platform? | Yes, given the same N and SD multiple | Depends on the reset point and on which volume the data contains |
The practical difference shows up in a trend. After a strong move, a 20-bar Bollinger middle line catches up within 20 bars and the bands reopen around the new level. A session VWAP keeps the memory of the whole morning, so price can stay outside the 2 SD VWAP band for hours. A rule that fades the Bollinger upper band and a rule that fades the VWAP upper band are therefore different rules with different trade counts, even on the same chart. If you have already tested one, the Bollinger Bands strategy backtest article shows the test structure; repeat it with the VWAP version and compare the two trade lists rather than assuming they agree.
Why does the session reset matter, and what is anchored VWAP?
The bands only mean something relative to where the average started, so a session VWAP that resets at each open and an anchored VWAP that starts at an event such as a settlement day or a gap give different levels on the same chart and must be tested as separate rules. Anchored VWAP is a VWAP whose running sums begin at a bar the trader chooses, for example the bar of an earnings release, a Federal Reserve announcement, a futures settlement day or the low of a sell-off, and then run forward across as many sessions as the trader wants.
For US index futures there are at least two common session resets: the exchange's evening open, which puts the whole overnight session into the average, and the cash-equity open, which starts the sums when stock trading begins. The overnight session carries much less volume than the cash hours, so the two VWAP lines are close by midday but far apart in the first hour. Decide which reset you trade, set the chart's session template to match, and feed the backtest the same bars. Exchanges can change session hours; the exchange's contract page is the place to confirm them.
Anchored VWAP is popular with traders who track dealer positioning around option expiries. The levels built from gamma exposure (GEX) are recalculated around the same event dates, and anchoring VWAP to the expiry bar gives a second reference that uses actual traded volume instead of open interest. Test the two apart from each other: an anchored band touch near a GEX level is a combined rule, and the combined rule needs its own test, with its own trade count, before you credit either part.
- Write down the reset or anchor rule in one sentence, for example "reset at the cash-equity open" or "anchor at the close of each monthly settlement day".
- Check that the bar data contains the volume the rule expects, including or excluding the overnight session.
- Run the band rule once per reset choice and keep the trade lists apart. If one choice works and the other does not, you have learned something about the reset, not about VWAP.
Which standard deviation should you use: 1, 2 or 3?
Width is a trade-off between how often price reaches the band and how far price has already stretched when it does, so 1 SD gives many small signals, 3 SD gives few large ones, and the choice is a parameter to test across a range rather than a fixed truth. If prices around VWAP followed a normal bell curve, about 32% of bars would sit beyond 1 SD, about 5% beyond 2 SD and under 1% beyond 3 SD. Intraday futures prices are not normal, and the cumulative calculation makes the early session wider, so the real shares differ by market and by time of day. The theoretical shares are still a useful map of the trade-off.
| Band width | How often price gets there | Move already made when it does | What it usually suits |
|---|---|---|---|
| 1 SD | Often, several times a session | Small | Exits, or entries with a second filter |
| 2 SD | A few times a session | Medium | The common first setting for reversion or trend rules |
| 3 SD | Rarely, not every day | Large | Event-driven rules; too few trades for a short test |
Test the multiple as a parameter, for example from 1.0 to 3.0 in steps of 0.25, and look at the whole row of results, not the single best cell. A rule that works at 2.0 and fails at 1.75 and 2.25 is fitting noise. A rule whose neighbours look similar is telling you something about the market. Pairing the band with a momentum filter, such as requiring the RSI indicator to read above 70 when price touches the upper band, cuts the trade count sharply; test the filtered and unfiltered versions side by side so you can see what the filter removed.
Band touch as mean reversion or band break as trend: which rule to test?
The same band supports two opposite rules, fading the touch back toward VWAP or following a close beyond the band, and the ES daily-bar run in the box, a stand-in VWAP rule that was the one candidate kept out of 1,024 tried for return against drawdown, shows what one such rule produced after costs without saying it will repeat. Mean reversion is the idea that a price stretched far from its average tends to return toward it; trend following is the idea that a price that has broken away keeps going. Both can be true on the same chart at different times of day, which is why neither is a property of the band.
| Rule element | Reversion rule | Trend rule |
|---|---|---|
| Entry | Sell short when price touches the +2 SD band (buy at −2 SD) | Buy when a bar closes above the +2 SD band (sell short below −2 SD) |
| Exit | At VWAP, or at the session close | On a close back inside the +1 SD band, or at the session close |
| Trades per day | More | Fewer |
| Typical trade | Small gain, occasional large loss when a trend continues | Small loss, occasional large gain when the move extends |
The run below comes from AlphaPrime (Valynox). It tested ES, the E-mini S&P 500 future, on daily bars from 2019-01-02 to 2025-06-30, long and short with one contract, with $20 commission and $25 slippage charged on every trade. This run is a stand-in: it uses a VWAP rule on daily bars as its only entry and exit, not an intraday band touch, because that is the run we have. The search tried 1,024 candidates and kept this one for its return against its drawdown; picking the kept one after the fact flatters it, in the way Bailey and colleagues describe for any best-of-many selection.
Source: AlphaPrime product engine (CPU), run 2026-10-31. ES daily bars, 2019-01-02 to 2025-06-30, long and short, 1 contract, $20 commission and $25 slippage per trade.
| Number | Result |
|---|---|
| Net profit (closed trades) | $109,660 |
| Max drawdown | $56,977.5 |
| Trades | 67 |
| Win rate | 65.67% |
| Profit factor | 1.769 |
| Average trade | $1,636.72 |
| Time in market | 91.62% |
Read the box as a description of one past path, not as what VWAP earns. Three numbers carry the information. Time in market of 91.62% means the rule was almost always holding a position and reversing, so a large part of the result is exposure to the index over a period that mostly rose. Sixty-seven trades over about six and a half years is a thin sample; a different six years could easily produce a different sign. The max drawdown of $56,977.5 against a net profit of $109,660 says that whoever traded one contract through this path sat through a loss of more than half the final gain. None of this says that a reader should expect these numbers, and nothing in the run tested the intraday touch rules in the table above; those need their own run on intraday bars.
How much do costs and slippage take from intraday VWAP band trades?
Intraday band trades capture small moves, so commission and one tick of slippage per side eat a large share of each trade, and the average trade after costs must be compared with the round-trip cost before the rule is judged. A reversion trade from the 2 SD band back to VWAP on a 5-minute chart is often a move of a handful of ticks. The round-trip cost is fixed in ticks regardless of how small the move is, so the smaller the target, the larger the share the cost takes.
Hypothetical example
A reversion rule on MES, the Micro E-mini S&P 500, captures 6 ticks per trade on average before costs. At $1.25 per tick that is $7.50. Commission and fees are $1.50 per round trip. Slippage is one tick on entry and one on exit, $2.50 in total. Round-trip cost is $4.00, more than half of the gross move. The net average trade is $3.50. If slippage rises to two ticks per side, the net falls to $1.00 per trade.
Tick and point values are set by the exchange and can change; the exchange's contract page is the place to confirm them, and the free futures contract calculator turns a move in ticks into dollars for the contract you trade. Run the band rule with the cost settings you would actually pay, then rerun it with double the slippage. A rule that survives both runs captured a real move; a rule that survives only the first was depending on perfect fills, which stop and market orders at a band touch rarely get. The number to read first is the average trade after costs: AlphaPrime's Performance Report shows it next to the trade count and the long and short split, so you can see in one view whether the short side at the upper band and the long side at the lower band are paying for their costs separately.
Your next step: take the reversion and trend rules from the table above, set the band multiple as a range from 1.0 to 3.0, charge at least one tick of slippage per side plus your commission, and run both rules on the same intraday bars with the same session reset. Then compare the two average trades with the round-trip cost before you look at any equity curve.
Frequently asked questions
Do VWAP bands work as support and resistance?
Sometimes price turns at a band and sometimes it runs through it, and a chart tends to show you the turns. The only way to know for your market and session is to write the touch as an exact rule, charge commission and slippage, and count the outcomes over several years. Test the reversion rule and the breakout rule separately, because the same touch belongs to both.
Is anchored VWAP better than session VWAP?
Each one answers its own question. Session VWAP tells you where today's volume traded; anchored VWAP tells you where volume has traded since a chosen event, which can be days or weeks ago. On the same bar they sit at different prices, so a rule built on one does not transfer to the other without a new test.
Can you use VWAP bands on daily bars?
You can, if the running sums start at a defined anchor, because a daily chart has no session open to reset at. The result is an anchored VWAP over many days, which behaves more like a slow moving average than an intraday level. The ES run in this article used a VWAP rule on daily bars as a stand-in, and its numbers say nothing about intraday band touches.
Which volume should VWAP use on a futures contract with a night session?
Use the volume that matches the reset you trade. If the chart resets at the exchange's evening open, the overnight volume is part of the average; if it resets at the cash-equity open, it is not. Mixing one choice on the chart with the other in the backtest produces bands at different prices and a trade list you cannot reproduce.
References
- Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
- Park, C.-H., & Irwin, S. H. (2007). What Do We Know About the Profitability of Technical Analysis? Journal of Economic Surveys, 21(4), 786–826.
- CME Group. FAQ: Micro E-mini Equity Index Futures. https://www.cmegroup.com/articles/faqs/frequently-asked-questions-micro-e-mini-equity-index-futures.html