VWAP

≈32 min read · Real gold futures examples with XAUUSD equivalents · Free

In this chapter: You will learn what the Volume-Weighted Average Price (VWAP) actually measures, why it began life as a yardstick for how well big institutions execute their orders, and how to calculate it by hand for a few gold candles. We will work through the standard deviation bands step by step, choose a session reset for gold deliberately rather than by accident, stack daily, weekly and monthly VWAPs, anchor VWAP to events, and read price location and VWAP slope as context. We finish by separating VWAP from moving averages and listing the seven most common VWAP mistakes. Throughout, the message is the same: VWAP is a reference for context, not a signal that tells you where price goes next.

In Chapter 3 you learned to count volume and delta: how many contracts traded, and which side was the aggressor (the side that crossed the spread with a market order). Those numbers live bar by bar. This chapter introduces the first tool that summarises all of that volume into a single, slowly evolving line on the chart. Candles show what happened to price; volume shows who, where and how hard. VWAP is the simplest way to fold the "where and how hard" into one number: the average price at which the market has actually done business.

What Is VWAP and Why Do Institutions Care?

VWAP (Volume-Weighted Average Price) is the average price at which trading took place over a defined period, where each price counts in proportion to the volume (number of contracts) that traded there. That second half of the definition is the whole point. A simple average treats every price the same. VWAP asks a sharper question: at what price did most of the actual business happen?

A tiny example makes the difference obvious. Suppose that in some short window 1,000 gold futures contracts trade at $4,200.0 and only 10 contracts trade at $4,210.0. The simple average of those two prices is $4,205.0, right in the middle. But almost nobody traded in the middle. VWAP weights each price by its volume:

VWAP = (4,200.0 × 1,000 + 4,210.0 × 10) ÷ (1,000 + 10) = 4,242,100 ÷ 1,010 ≈ 4,200.1

So VWAP sits almost exactly where the business was done. The 10 contracts at $4,210 barely move it, which is exactly what you would want from a measure of "the typical traded price".

Born as an execution benchmark

VWAP was not invented for retail chart reading. It became popular as a benchmark for execution quality, that is, how well (how cheaply or expensively) a large order was filled. A well-known 1988 paper by Berkowitz, Logue and Noser in the Journal of Finance, "The Total Cost of Transactions on the NYSE", used the day's VWAP as the reference against which institutional trades were measured. The logic is simple. If a pension fund buys a large block of shares throughout the day and its average purchase price is below that day's VWAP, it bought more cheaply than the average participant, and its trading desk is judged to have executed well. If it paid above VWAP, it did worse than average.

That idea is still everywhere. Many large orders today are handled by VWAP algorithms (often called "VWAP algos"): programs that slice a big order into many small pieces and release them in proportion to the market's typical volume pattern, so that the final average fill lands close to the day's VWAP. For a desk that has to buy 2,000 contracts without moving the market, "match VWAP" is a reasonable, measurable goal.

Gold note: Gold futures have their own built-in connection to VWAP. Under CME's settlement procedures, the daily settlement price of the active GC contract is based on the volume-weighted average price of trades in a short window, 13:29–13:30 New York time. In other words, the exchange itself uses VWAP logic to decide the official daily price. Procedures can change, so check CME's current settlement documentation before relying on the exact details.

What VWAP tells you, and what it does not

At any moment during the session, VWAP tells you where the average traded price sits so far, counting from its starting point. That is context: it tells you whether the current price is expensive or cheap relative to the business done today. If gold trades at $4,212 and VWAP is $4,203, current trades are happening well above the day's average transaction price. That is a fact about the present and the past. It is not a forecast.

Two features of VWAP will matter in every section that follows:

• VWAP is cumulative. It includes every trade from its start point (called the reset or anchor) up to now. Nothing ever drops out.
• VWAP only has its precise meaning on real exchange volume. Futures such as GC (100 troy ounces) and MGC (10 troy ounces) report every contract traded on one central exchange. Most CFD (contract for difference) brokers offering XAUUSD show only tick volume, a count of how many times the broker's own price feed changed. Tick volume is not the number of contracts that changed hands, so a "VWAP" built from it is a rough, broker-specific approximation.

Key idea: VWAP is the market's average traded price, weighted by volume, measured from a chosen start point. It began as a benchmark for grading executions, not as a prediction tool.

Common mistake: "Institutions always defend VWAP." This is a popular claim, not proven. Being a benchmark means fills are compared against VWAP after the fact; it does not mean anyone is obliged to buy or sell there. A related claim, "price always returns to VWAP like a magnet", is also a popular claim, not proven. On strong trend days price may never come back to it.

The VWAP Formula, Step by Step

The general formula is short:

VWAP = Σ(Price × Volume) ÷ Σ(Volume)

The symbol Σ (the Greek capital letter sigma) simply means "add up". So: multiply each traded price by the volume traded at that price, add all those products together, and divide by the total volume.

If you had every single trade, you would apply this directly to each one. On a candle chart, however, the platform usually knows only the open, high, low, close and volume of each candle. So it picks one representative price per candle. The most common choice is the Typical Price (TP):

TP = (High + Low + Close) ÷ 3

With that, the calculation becomes five mechanical steps:

1. Compute the typical price of each candle.
2. Multiply the typical price by that candle's volume (TP × V).
3. Keep a running total of TP × V.
4. Keep a running total of volume.
5. Divide the running total of TP × V by the running total of volume.

A worked example on micro gold

Here are four one-minute MGC candles. The prices are simplified and hypothetical so the arithmetic is easy to follow; real gold prices will differ.

Check one row yourself. Candle 3: TP = (4206 + 4202 + 4204) ÷ 3 = 4204. Multiply by 700 contracts: 2,942,800. Add to the previous running total of 840,200: 3,783,000. Running volume: 200 + 700 = 900. Divide: 3,783,000 ÷ 900 = 4,203.33.

Now compare two "averages" of these four candles:

• The simple average of the four typical prices is (4200 + 4202 + 4204 + 4201) ÷ 4 = 4201.75.
• The VWAP after candle 4 is 4203.10.

Why the difference of more than a dollar? Candle 3 carried 700 of the 1,000 contracts, which is 70% of all the volume. VWAP is pulled towards where the volume was. That is what "volume-weighted" means in practice: the heavy candle wins.

Gold note: These are hypothetical futures prices. If you wanted to mark the same level on a spot-gold CFD chart, you would subtract the current futures-to-spot gap, which has recently been roughly $27–30 (always approximate, and it changes over time). A GC VWAP of 4203.1 would sit at roughly 4173–4176 on XAUUSD. Read the live gap at the time; never hard-code it.

Why VWAP slows down as the day goes on

Look again at candle 4. Price dropped back to a typical price of 4201, more than two dollars below the VWAP. Yet VWAP only moved from 4203.33 to 4203.10, about 23 cents. The reason is that 100 new contracts are being averaged against 900 old ones.

You can see this precisely with an equivalent "update" form of the formula:

New VWAP = Old VWAP + V_new × (P_new − Old VWAP) ÷ (Old Σ V + V_new)

For candle 4: 4203.33 + 100 × (4201 − 4203.33) ÷ (900 + 100) = 4203.33 − 0.233 ≈ 4203.10.

The denominator keeps growing all session. Early in the day a single busy candle can shove VWAP around; late in the day the same candle barely nudges it. This is not a flaw. It is simply what a cumulative average does. A trader who complains that "VWAP is too slow in the afternoon" is really observing that the afternoon VWAP already contains most of the day's business.

A few practical consequences:

• Dividing by the number of candles instead of the total volume is the most common hand-calculation error. That gives you a simple average, not VWAP.
• VWAP is expressed in price units: dollars per troy ounce for gold.
• Two platforms fed with identical data and identical settings should show practically the same VWAP. When they differ, the cause is usually a settings difference (session start, price source, data resolution), which the next two sections cover.

Try it: Candle A trades at a price of 100 with a volume of 900. Candle B trades at 110 with a volume of 100. Before calculating, guess whether VWAP is closer to 100 or 110. Then compute it: (100 × 900 + 110 × 100) ÷ 1,000 = 101. The simple average would be 105.

Which Price Goes Into VWAP? Typical Price vs Tick Data

The "ideal" VWAP is computed from every individual trade: the price of each trade multiplied by its size, summed and divided by the total number of contracts. NinjaTrader's documentation for its order-flow VWAP describes it in exactly these terms: total dollars traded divided by total contracts traded. We will call this tick-level VWAP, because it is built from tick-by-tick trade data, the raw record of each transaction.

Most charts, though, are built from candles, and a candle does not record where inside its range each contract traded. So the platform chooses a single source price for each candle and pretends the candle's entire volume traded there. Common choices:

Every one of these is an assumption: that the candle's volume was concentrated at one price. On a one-minute candle in a quiet market the error is tiny, because the candle's whole range may be only a few ticks. On a 30-minute candle during a news release, the error can be large.

How big can the error be?

Consider one 30-minute GC candle on a news day (hypothetical numbers): open 4201, high 4220, low 4200, close 4218, with 1,000 contracts traded. Suppose that, inside the candle, the volume was distributed like this:

Most of the business happened low in the candle, around 4202–4204, before the release; then price jumped and traded less. The true, trade-by-trade VWAP of this candle is:

(4200×20 + 4202×290 + 4204×290 + 4216×190 + 4218×190 + 4220×20) ÷ 1,000 = 4208.6

Now compare what each candle-based source would assume:

None of the shortcuts "knows" that most contracts traded near the bottom of the candle. On this one candle the Close method is off by more than nine dollars, about 94 ticks. On one-minute candles the same news move would be split into many small candles, each with a much narrower range, and the approximation error would shrink dramatically.

Resolution: calculate fine, view coarse

A second setting is resolution: what data the platform uses underneath the chart you are looking at. Some platforms, NinjaTrader among them, can compute VWAP from one-tick data even while you view a five-minute chart. That is more accurate but heavier on your computer.

A practical rule for intraday gold work: calculate VWAP on the finest data you can (tick data or one-minute bars), even if you look at a five-minute or fifteen-minute chart.

Key idea: When your VWAP and someone else's VWAP disagree by a few ticks, neither is necessarily "wrong". Check three settings first: session (reset time), source price, and data resolution.

Common mistake: Believing that choosing HLC3 over Close, or vice versa, gives you an "edge". On fine data the difference is usually small. The source setting is about accuracy and consistency, not about secret advantages. Equally, calculating VWAP from daily candles and expecting intraday precision is a mismatch of tools.

When Does Gold's VWAP Reset? Globex vs RTH

A daily VWAP needs a starting point, the moment the running totals are set back to zero. For US stocks this is easy: the regular session opens at 09:30 New York time and that is the obvious reset.

Gold futures are different. As covered in Chapter 1, GC and MGC trade on CME Globex, the exchange's electronic platform, from Sunday 18:00 to Friday 17:00 New York time (ET), with a daily halt from 17:00 to 18:00 ET. CME's trading day for gold begins at 18:00 ET on the previous evening. That leaves two common choices for the reset.

Option 1: Globex reset (18:00 ET)

The VWAP starts at 18:00 ET and includes everything: the Asian session, London, and New York.

• Advantage: it lines up with CME's official trading day and leaves no volume out.
• Disadvantage: by the New York morning, a large amount of overnight volume is already baked in. If you mainly care about how US participants are pricing gold, the overnight weight can make VWAP less representative of the "main" session.

Option 2: RTH reset (about 08:20 ET)

Many platforms define a Regular Trading Hours (RTH) session for COMEX metals based on the old trading-floor ("pit") hours, roughly 08:20 to 13:30 ET. The rest of the day is then called the Electronic Trading Hours (ETH) or overnight session.

• Advantage: it focuses on the hours with heavy US participation, the usual time of major US economic releases (often 08:30 ET), and the daily settlement window (13:29–13:30 ET).
• Disadvantage: it ignores overnight volume, and sometimes the most important move of the day happened overnight in London.

Neither choice is "more correct". They answer different questions. A Globex-reset VWAP answers "what is the average traded price of CME's whole trading day?" An RTH-reset VWAP answers "what is the average traded price during the main US hours?" Some traders display both at once, one for the overnight and full session and one for RTH.

Our rule throughout this book is short: pick one, label it on the chart, and stay consistent. A VWAP without a session label cannot be compared with anything.

Gold note: Session times are quoted in New York time. When you convert to your local time zone, remember that daylight saving time changes on different dates in different countries, so the local equivalent of 18:00 ET can shift by an hour for a few weeks each spring and autumn. Always check CME's official hours before relying on them.

Common mistake: Comparing your VWAP with a screenshot from a video or a colleague without knowing which session they used. Another frequent one is assuming gold "opens" at 09:30 ET like stocks; it has been trading for many hours by then.

Try it: Put two VWAPs on the same five-minute GC chart, one reset at 18:00 ET and one at 08:20 ET. At 10:00 ET, write down the distance between them. Do this for five days and note on which days the overnight session pulled the Globex VWAP far from the RTH one.

How VWAP Standard Deviation Bands Are Calculated

Most VWAP indicators can draw bands above and below the VWAP line. To understand them you need one statistical idea.

The standard deviation, written σ (lower-case sigma), measures how far values typically sit from their average. A small σ means prices clustered tightly around VWAP; a large σ means they were spread out. The variance is σ squared: the average of the squared distances from the mean. We square the distances so that values above and below the mean do not cancel each other out, and so that far-away values count more heavily.

For VWAP, the common and logical version is volume-weighted. Sierra Chart documents it in this form:

σ² = Σ[ V × (TP − VWAP)² ] ÷ Σ V, and σ = √σ²

An equivalent shortcut, handy in a spreadsheet, is σ² = Σ(V × TP²) ÷ Σ V − VWAP².

The bands are then:

Upper band k = VWAP + k × σ, and Lower band k = VWAP − k × σ

where k is usually 1 and 2, sometimes 3.

Worked example: continuing the four candles

From the previous example, VWAP = 4203.10 and total volume = 1,000.

σ² = 2,090 ÷ 1,000 = 2.09, so σ = √2.09 ≈ $1.45.

The bands, rounded to gold's tick of $0.10:

Notice something about candle 1. It had only 100 contracts, the same as candles 2 and 4, yet it contributes 961 of the 2,090 total, nearly half. That is because it sat furthest from VWAP and the distance is squared. Prices far from the average have an outsized effect on σ.

Bands are cumulative too, and noisy early

Like VWAP, σ is computed from the reset onward. Watch how it evolved in our four-candle example:

After the first candle, with only one typical price, σ is zero and the bands collapse onto VWAP. With very little data, every new candle changes σ noticeably. As the session progresses and volume accumulates, the bands usually widen and become steadier. In practice, the bands in the first fifteen to thirty minutes after a reset are unreliable simply because they are built on too little information.

Common mistake: Thinking the ±1σ band is a fixed dollar distance. It changes every day and every minute. A related mistake is comparing bands from two platforms that use different formulas: some use a volume-weighted σ, some a plain standard deviation, and some (Sierra Chart among them) also offer percentage or fixed-distance bands. Read your platform's documentation before comparing.

What VWAP Bands Do (and Don't) Tell You

Here is where many traders go wrong, so it is worth slowing down.

In statistics, if data follow a normal distribution (the familiar symmetric "bell curve"), about 68% of values lie within ±1σ of the mean and about 95% within ±2σ. This is called the 68–95 rule. Many traders transfer it straight onto VWAP bands: "Price is at +2σ, so there is a 95% chance it reverses." That conclusion is wrong, for three separate reasons.

Reason 1: prices are not normally distributed

Market returns have fat tails: very large moves happen far more often than a normal distribution would predict. Benoit Mandelbrot made this point famously (see The (Mis)Behavior of Markets, with Richard Hudson, 2004). In gold, the tails show up most clearly around major US releases such as CPI (inflation), NFP (the monthly jobs report) and FOMC (the Federal Reserve's interest-rate decisions). On those days, "three-sigma" or "four-sigma" moves relative to the morning's bands are not rare at all.

Reason 2: σ describes the past, not the future

The band tells you, "so far today, trades have typically been this far from the average." If the market changes character, for example when a news release arrives, σ adapts only slowly, because it is cumulative. Even the statement "95% of today's volume traded inside ±2σ" would be a description of trades that already happened in this session, not a probability for the next trade.

Reason 3: the bands move with price

VWAP and σ are both cumulative and both respond to new trades. On a strong trend day, price can ride the +1σ or +2σ band for hours: price climbs, new volume trades at higher prices, VWAP rises, σ widens, and the band climbs right behind price. The band is not a wall; it is attached to the thing it is measuring.

So what are the bands good for?

Bands are a normalised ruler for stretch. Saying "price is $6 above VWAP" means different things on different days. On a quiet day with σ = $3, $6 above VWAP is 2σ, a large stretch relative to that day's activity. On a CPI day with σ = $12, $6 above VWAP is only 0.5σ, barely noteworthy. Expressing distance in σ units lets you compare across days and volatility regimes:

Distance in σ = (Price − VWAP) ÷ σ

Using our worked example, if price were 4205.0 with VWAP 4203.1 and σ 1.45, the distance would be (4205.0 − 4203.1) ÷ 1.45 ≈ 1.3σ.

That is a useful description. Whether anything happens at that stretch is a separate question, answered by other evidence: the order-flow behaviour at that price (absorption, aggression, delta, covered in Chapters 3, 8 and 9), the structure of the day, and the volume profile (Chapter 5).

Key idea: "+2σ" honestly means "stretched relative to today's traded distribution so far." It does not mean "95% likely to reverse."

Common mistake: "±2σ means a 95% chance of reversal" is a popular claim, not proven, and it misapplies the statistics. So is treating ±1σ as an automatic buy or sell line. Treating every touch of +2σ as a turning point on a trend day is a classic way to be repeatedly on the wrong side of a band ride.

Try it: Pick two recent GC days: one quiet, range-bound day and one major-news day. On each, note the σ value at 11:00 ET and how many times price closed a five-minute bar outside ±2σ. You will see how differently the same band setting behaves.

Daily, Weekly and Monthly VWAP

The VWAP formula never changes. Only the reset point changes:

• Daily VWAP: from the start of the session (Globex 18:00 ET, or RTH about 08:20 ET).
• Weekly VWAP: from the start of the trading week; for gold, Sunday 18:00 ET.
• Monthly VWAP: from the first trading session of the month.

Each one reflects a different time horizon of participants. The daily VWAP is the average traded price of today's business, dominated by intraday traders. The weekly and monthly VWAPs average much longer stretches, which include the volume of slower participants: swing traders who hold for days, funds that build positions over weeks, and commercial hedgers such as miners and jewellers.

Stacking context

Putting all three on one chart gives you a layered description. Suppose gold is trading above the daily VWAP but below both the weekly and the monthly VWAP. A careful, descriptive reading is: today, business is being done above today's average price, but on the weekly and monthly horizons price is still below the average transaction price. That is not a signal to do anything. It is a reminder of which bigger picture today's activity sits inside.

Why the monthly VWAP goes flat

By late in the month, the monthly VWAP contains an enormous amount of volume, so each new bar has almost no weight. Using illustrative round numbers and the update formula from earlier:

• Daily VWAP at 08:30 ET with, say, 40,000 contracts traded so far. A five-minute bar of 2,000 contracts trades $10 above it. The VWAP moves by 2,000 × 10 ÷ 42,000 ≈ $0.48.
• Monthly VWAP on the twentieth trading day with, say, 4,000,000 contracts so far. The same bar moves it by 2,000 × 10 ÷ 4,002,000 ≈ $0.005, less than a tenth of one tick.

So a nearly flat monthly VWAP late in the month is completely normal.

Watch out for the contract roll

Futures contracts expire, and traders move their positions from the expiring contract to the next one, a process called the roll or rollover. Gold's main active months are usually February, April, June, August, October and December, and volume migrates to the next active contract towards the end of each cycle. A weekly and especially a monthly VWAP can straddle a roll.

How this affects you depends on your price series:

• A continuous back-adjusted series shifts historical prices so the chart has no jump at the roll. Old prices on the chart are then not the prices that actually traded.
• A non-adjusted series may show a sudden jump equal to the price difference (the spread) between the two contract months.

Either way, a long VWAP that crosses a roll needs care. Always know which price series your weekly or monthly VWAP is calculated on.

Common mistake: "The monthly VWAP is guaranteed support or resistance" is a popular claim, not proven. Also common: comparing a monthly VWAP computed on two different contract months, and covering the chart with ten VWAPs without a clear question for each.

Anchored VWAP: Start the Clock at an Event

An Anchored VWAP (AVWAP) is an ordinary VWAP whose starting point you choose, instead of letting the calendar choose it. The formula is unchanged: Σ(TP × V) ÷ Σ V from the anchor candle up to now. Brian Shannon of Alphatrends popularised the approach (see his book Maximum Trading Gains with Anchored VWAP, 2023). His central idea is that the average traded price is most meaningful when measured from a significant event, not from an arbitrary clock time.

Common anchors for gold:

• News events: the CPI or NFP release (usually 08:30 ET) or the FOMC statement (usually 14:00 ET).
• Significant swing highs or lows: points where the market clearly changed direction.
• Session or week starts: which simply reproduce the daily or weekly VWAP.
• Unusual days: an exceptionally high-volume day, or a price gap after a weekend or holiday.

What an AVWAP measures

An AVWAP anchored at an important low is the average price of every contract traded from that low until now. Shannon's common interpretation is that above a rising AVWAP, the participants who have traded since that event are, on average, in profit, and below a falling AVWAP the reverse.

That interpretation is a useful simplification, but be precise about its limits. Every contract has a buyer and a seller. VWAP is the average price of trades, not the average cost of open positions. Some of those trades opened positions, some closed them, and the buyer of each contract had a seller who now holds the opposite result. "Who is in profit" is a helpful mental picture, not a measurement.

The weak spot: choosing the anchor

The main weakness of AVWAP is that the choice of anchor is subjective, and Shannon himself acknowledges this. The danger is hindsight bias: after the fact, it is easy to pick the anchor that "worked best" and then believe the tool works. If you look at a chart, try five anchors, and keep the one whose line price happened to respect, you have learned nothing about the future.

The remedy is a rule written before you look: for example, "always anchor at the 08:30 ET candle on CPI days" or "always anchor at the low of the prior week". Then apply that rule every time, including on the days it looks unimpressive.

Key idea: Anchored VWAP is the same formula with a deliberately chosen window. Its value depends entirely on choosing that window by a rule, not by hindsight.

Common mistake: "Price will definitely return to the AVWAP from the last low" is a popular claim, not proven. Another frequent error is anchoring to insignificant candles until the chart is a tangle of lines.

Try it: Write down one anchor rule today, for example "the 14:00 ET candle on FOMC days". Apply it to the last six FOMC days on a 15-minute GC chart without skipping any. Note honestly what you see on each.

Price vs VWAP and VWAP Slope as Context

VWAP gives you two simple, descriptive pieces of information.

1. Location

If price is above VWAP, current trades are happening at prices more expensive than the average traded price since the reset. Below VWAP means cheaper. When price stays above VWAP for a long time, the market is showing acceptance of higher prices: it is willing to keep doing business up there. (Acceptance will be defined more carefully with volume profiles in Chapters 5 to 7.)

2. Slope

Because VWAP is cumulative, its slope tells you where new business is being done relative to the old average. A rising VWAP means new trades keep happening above the existing average; a falling VWAP means they keep happening below it. A flat VWAP means trading is rotating around one fair price, which is a sign of balance: buyers and sellers broadly agree on value and the market is not moving away from it.

A simple context matrix

Combining location and slope gives a simple descriptive grid:

Shannon puts it simply: above a rising VWAP, buyers are in control until proven otherwise. We treat this as context. Its job is to help you recognise whether the market is in a trending state or a balanced one, so that you know which tools and which questions are appropriate. It is not an entry rule.

Add the σ ruler from earlier. "Price is above VWAP by 0.3σ" and "price is above VWAP by 2.5σ" are very different stories, even though both are "above VWAP".

Be cautious with slope early in the session. In the first minutes after a reset, VWAP is built on very little volume and its slope can swing wildly. Also remember that the slope depends on the session you chose: an RTH VWAP and a Globex VWAP on the same morning can slope in different directions.

Common mistake: "Above VWAP, only buy; below VWAP, only sell." This is a popular claim, not proven: a simplification that cannot replace context. Reading slope in the first ten minutes after the reset is another frequent error.

Try it: Find three recent GC days on a five-minute chart with an RTH VWAP: one that trended up, one that trended down, and one that rotated. Count how many times price crossed VWAP on each day and describe each day using the matrix above.

VWAP vs Moving Average: Not the Same Line

VWAP and a moving average are both "averages" and can look similar on a chart. They answer different questions.

A moving average (MA) is the average of the last N closing prices. A Simple Moving Average (SMA), such as the SMA 20, gives each of the last 20 candles equal weight. An Exponential Moving Average (EMA) gives more weight to recent candles and less to older ones. There are three structural differences between these and VWAP.

1. The window

A moving average uses a rolling window: the SMA 20 always averages the latest 20 candles; as a new candle arrives, the oldest one drops out. VWAP uses a cumulative window: everything from the reset or anchor up to now, with nothing ever dropping out.

2. The weighting

The standard SMA weights all candles equally; the EMA weights recent candles more. Neither looks at volume in its standard form. VWAP weights by volume: a candle with 5,000 contracts has fifty times the influence of a candle with 100 contracts.

3. The economic meaning

VWAP is a real quantity: the average price at which contracts actually changed hands since the reset, the very number a fund's trading desk uses to grade its executions. A moving average is a statistical smoothing tool. The "20" or "50" in its name is a convention among traders, not a property of the market.

The practical consequence: a 20-period EMA reacts to a sharp 11:00 ET move at roughly the same speed it would at 09:00 ET. VWAP reacts strongly at 09:00 (little accumulated volume) and only slightly at 12:00 (lots of accumulated volume). Neither is "better". VWAP is the right tool for the question "where is price relative to today's business?"

There is also a hybrid called Rolling VWAP: a VWAP calculated over a rolling window of the last N candles. It weights by volume like VWAP but forgets old data like a moving average.

Common mistake: "VWAP is just an EMA in a different colour." It is not; the window, weighting and meaning are all different. Expecting VWAP to react at noon as quickly as it did at 9 a.m. is another frequent misunderstanding. And "the 200 MA or VWAP is a magic line" is a popular claim, not proven.

7 Common VWAP Mistakes

Here is the chapter condensed into seven mistakes, each with the section where it was explained.

1. Not knowing the reset session. A VWAP reset at 18:00 ET and one reset at 08:20 ET are two different numbers. Decide and label your session first (see "When Does Gold's VWAP Reset?").

2. Using data without real volume. On most CFD platforms, XAUUSD "volume" is tick volume: the number of price changes in that broker's own feed, not centralised exchange volume. A VWAP built on it is approximate and depends on the broker. GC and MGC futures report real exchange volume. If you trade XAUUSD, a sounder approach is to read VWAP on the futures and translate the level using the current futures-to-spot gap (recently roughly $27–30, always approximate).

3. Treating VWAP as a magnet. "Price always returns to VWAP" is a popular claim, not proven. On trend days it may never return.

4. Treating bands as probabilities. ±2σ is not a 95% chance of reversal (see "What VWAP Bands Do (and Don't) Tell You").

5. Trusting VWAP and bands in the first minutes. With little cumulative volume, every candle shifts VWAP and σ noticeably.

6. Ignoring the contract roll on weekly and monthly VWAPs (see "Daily, Weekly and Monthly VWAP").

7. Replacing context with a signal. VWAP is a reference. Price reaching VWAP means only that price has reached today's average traded price. What happens there, whether aggressive orders are absorbed, whether delta shifts, whether the structure holds, has to be read with order-flow tools.

There is an eighth mistake that affects learning more than trading: selection bias in examples. If you only ever look at, or are only ever shown, days when price bounced neatly off VWAP, you will build a false picture. For every clean example you study, find one where the same idea failed, using the same settings. This book tries to follow that rule, and you should demand it from any educational material.

Your VWAP setup checklist

Common mistake: "VWAP alone is a complete strategy" and "institutions hide orders at VWAP" are both popular claims, not proven. Comparing a CFD gold VWAP directly with a futures VWAP is also a mismatch: different instruments, different volume, and a price gap between them.

Key idea: VWAP is more information about where business has been done. It is not a magic edge. Our own testing of order-flow information in general found that signals on their own did not reliably predict direction once trading costs were included, which is why this book treats every tool as context to be combined, not as a trigger.

Chapter summary

• VWAP is the volume-weighted average traded price from a chosen start point: Σ(Price × Volume) ÷ Σ Volume.
• It began as an institutional execution benchmark (Berkowitz, Logue and Noser, 1988), and CME uses a short VWAP window (13:29–13:30 ET) in settling the active GC contract.
• On candle charts VWAP uses a source price per candle, most often the typical price (H + L + C) ÷ 3; this is an approximation of the true tick-level VWAP, and the error grows with candle size.
• VWAP is cumulative: it is responsive early in the session and slow later, by design.
• For gold you must choose a reset: Globex (18:00 ET, the full CME day) or RTH (about 08:20 ET, the main US hours). Label it and stay consistent.
• Standard deviation bands are VWAP ± k × σ, with σ usually volume-weighted. They are cumulative, noisy early, and implemented differently across platforms.
• The 68–95 rule does not turn bands into reversal probabilities: prices have fat tails, σ describes the past, and bands move with price. Use bands as a ruler of stretch.
• Daily, weekly and monthly VWAPs share one formula with different resets; together they layer context. Long VWAPs can be distorted by contract rolls.
• Anchored VWAP starts the clock at an event; choose anchors by a written rule to avoid hindsight bias.
• Location (above or below) and slope (rising, flat, falling) describe trending versus balanced context, not entries.
• VWAP differs from moving averages in window, weighting and economic meaning.

Checklist

• I can define VWAP in one sentence and explain why it is a benchmark rather than a forecast.
• I can calculate VWAP by hand for a few candles using typical price and running totals.
• I know my platform's VWAP source price and data resolution.
• I have chosen a session reset for gold (Globex or RTH) and labelled it on my chart.
• I can calculate a volume-weighted σ and the ±1σ and ±2σ bands for a small example.
• I describe price distance from VWAP in σ units, not just dollars.
• I do not treat ±2σ as a 95% reversal probability.
• I know which contract or continuous series my weekly and monthly VWAPs use, and when the last roll was.
• If I use anchored VWAP, my anchor rule is written down before I look at the chart.
• I use VWAP on real exchange volume (GC/MGC) and translate levels to XAUUSD with the live, approximate gap.

Quiz

1. Candle A trades at a price of 100 with a volume of 900. Candle B trades at 110 with a volume of 100. What is the VWAP of the two candles, and what is their simple average?
2. Why does a session VWAP move much less in response to a busy candle at 12:00 ET than to the same candle at 08:40 ET?
3. VWAP is 4203.1 and the volume-weighted σ is 1.45. Where are the −2σ band and the +1σ band (rounded to the tick)?
4. Price is at +2σ above VWAP. Which statement is honest: (a) there is a 95% chance price reverses; (b) price is stretched relative to today's traded distribution so far; (c) sellers must step in here?
5. Name two reasons why a VWAP drawn on a spot-gold CFD chart is less reliable than one drawn on GC futures.

Quiz answers

1. VWAP = (100 × 900 + 110 × 100) ÷ 1,000 = 101,000 ÷ 1,000 = 101. The simple average is (100 + 110) ÷ 2 = 105. VWAP sits near the price where most of the volume traded.
2. Because VWAP is cumulative. At 12:00 ET the running total of volume is much larger, so the new candle's volume is a much smaller fraction of the whole and its weight in the average is smaller. The update formula shows this: the change equals V_new × (P_new − VWAP) ÷ (total volume including the new candle).
3. −2σ = 4203.1 − 2 × 1.45 ≈ 4200.2; +1σ = 4203.1 + 1.45 ≈ 4204.5 (approximately 4170–4177 on XAUUSD with a gap of roughly $27–30).
4. (b). The band describes stretch relative to the trades so far today. Price distributions have fat tails, σ describes the past, and bands move with price, so (a) misapplies the 68–95 rule; (c) is an unsupported claim.
5. Any two of: CFD "volume" is usually broker tick volume (price-change counts), not centralised exchange volume; it is specific to one broker's feed, so two brokers give different VWAPs; the CFD price differs from the futures price by a moving gap, so CFD and futures VWAPs cannot be compared directly.

Education only. Not financial advice. Trading involves substantial risk of loss.