VWAP Indicator: Formula, Anchors, and What Tests Reveal

VWAP, short for Volume Weighted Average Price, is an average price that gives more weight to observations with more volume. Traders use it as an execution benchmark and a reference for reading the current price.

Our 19,392 configuration evaluations compare volume weighting, price sources, reset anchors, and simple trading rules. One useful distinction appears before any performance ranking: at the start of a new anchor, a candle-based VWAP may describe only that candle’s own price position. The volume has no influence until another observation joins it.

What does VWAP show?

A price above VWAP sits above the volume-weighted average since the chosen starting point. A price below it sits below that average. Neither relationship establishes where the next price will go.

Schwab’s explanation describes VWAP’s familiar intraday use. Execution benchmarking and a directional trading signal are different applications: buying below an average can improve an execution comparison while the market continues falling.

The anchor says when the calculation starts again. Session VWAP commonly resets each trading day. Other versions start at a week, a month, or a selected event. Changing that starting point changes which prices and volumes the line represents.

How to calculate VWAP

The underlying idea is:

VWAP = sum of price × volume since the anchor
       ÷ sum of volume since the anchor

Transaction-level VWAP can use individual trade prices and quantities. With candles, a common approximation supplies one price per candle, such as typical price = (high + low + close) / 3, and weights it by that candle’s volume. Our tests use this bar-based calculation and compare it with closing-price input.

For two observations priced at 100 and 110, volumes of 9 and 1 give a VWAP of 101. Equal weights give an average of 105. The observations are unchanged; their influence differs.

TradingView’s VWAP documentation explains the cumulative candle calculation, price-source setting, and reset choices. Source and anchor belong in a reproducible VWAP description.

Why the first bar needs a different interpretation

Immediately after a reset, the bar-based formula has only one observation. If its volume is positive:

First-bar VWAP = source price × volume ÷ volume
               = source price

With close as the source, VWAP equals the close. A strict close-above-VWAP rule cannot qualify on that first bar.

With typical price as the source, close above VWAP simplifies to:

close > (high + low + close) / 3

which is equivalent to

close > (high + low) / 2

The signal asks whether the candle closed above the midpoint of its high-low range. Volume cancels from both versions. This is an algebraic property of the one-observation candle approximation; it does not describe the average of every transaction inside that candle.

The script preserves the one-observation equality explicitly at each reset, preventing rounding from turning a tie into a strict above-or-below signal. Historical candle checks also reproduced the typical-price midpoint relation. An early reading can be useful, but its meaning changes as more observations accumulate.

Equality also matters for exits. A close-source rule that exits at or above VWAP can close on the first reset bar. A rule that requires the close to be strictly below VWAP does not qualify on that equality. The tests keep those definitions fixed.

Can volume weighting change a signal?

Yes. An equal-bar average and VWAP can place the same close on opposite sides of the line even when they share the same source and anchor.

Select Same close, different side. Both averages reset together. Higher-volume observations pull the blue line more strongly, while every observation has equal influence on the amber line.

This selected episode illustrates the calculation. With daily anchors and closing-price input, strict opposite price-to-average signs appeared on about 1.8–3.2% of checked candles across the historical samples. Volume can change the reading even when the anchor and prices stay fixed. That frequency does not establish which line provides a better future forecast.

How anchors and buffers change trading activity

We tested two explicit rules. A strength rule buys above an average plus a buffer and exits below the average. A fade rule buys below the average minus a buffer and exits at or above it.

Additional tests using small positions hold the source, weekly anchor, and strength rule fixed. Each point in the next chart changes only the required percentage distance from the average.

The direction of the weighting difference also changes with the threshold. In the illustrated sample, a 0.50% buffer recorded 704 executions with volume weights and 674 with equal bars. At a 2% buffer, the counts were 320 and 322.

A wider buffer changes which entries qualify, when a position starts, and which later signals arrive while that position is open. Execution counts describe the resulting strategy path, not every candle above the line.

Daily and monthly anchors represent different amounts of history. Calling one a better VWAP setting would conceal that change in the question. The complete fixed-grid comparisons, including returns, drawdowns, and inactive settings, remain in the evidence.

VWAP versus VWMA, and useful settings

VWMA, a volume-weighted moving average, normally uses a rolling number of bars. Its oldest observation leaves whenever a new one arrives. Anchored VWAP keeps accumulating until its reset. A 20-bar VWMA and a daily VWAP can therefore contain different observations even on the same chart.

Choose a timeframe with enough observations inside the anchor to answer the intended question. A session anchor on daily candles leaves one candle per session; the bar approximation then collapses to its source every bar. Our relatively coarse intraday tests answer questions about those candles, not minute-by-minute execution quality.

There is no universal best source, anchor, or buffer. A useful comparison holds two of them fixed and changes the third, then repeats the same rule on another historical sample.

Reproduce the comparison

The script defaults to daily anchors, closing-price input, volume weighting, and an unbuffered strength rule. Switch the weight to equal bars to isolate the effect of volume.

//@version=6
strategy("VWAP Lab: anchors, weighting and buffers", overlay=true, initial_capital=10000, default_qty_type=strategy.percent_of_equity, default_qty_value=99, process_orders_on_close=false)

anchor_mode = input.int(0, "UTC anchor: 0 day, 1 week, 2 month", minval=0, maxval=2)
source_mode = input.int(0, "Source: 0 close, 1 typical price", minval=0, maxval=1)
weight_mode = input.int(0, "Weight: 0 volume, 1 equal bars", minval=0, maxval=1)
rule_mode = input.int(0, "Rule: 0 strength, 1 fade", minval=0, maxval=1)
buffer_pct = input.float(0.0, "Entry buffer percent", minval=0.0, maxval=5.0)

source_value = source_mode == 0 ? close : hlc3
day_reset = timeframe.change("1D")
week_reset = timeframe.change("1W")
month_reset = timeframe.change("1M")
is_reset = bar_index == 0 or (anchor_mode == 0 ? day_reset : anchor_mode == 1 ? week_reset : month_reset)
bar_weight = weight_mode == 0 ? volume : 1.0
var float weighted_total = 0.0
var float weight_total = 0.0
weighted_total := is_reset ? source_value * bar_weight : weighted_total + source_value * bar_weight
weight_total := is_reset ? bar_weight : weight_total + bar_weight
// Preserve the one-observation identity exactly at each reset.
anchor_mean = weight_total > 0 ? (is_reset ? source_value : weighted_total / weight_total) : na
ready = bar_index >= 500 and not na(anchor_mean)
entry_level = anchor_mean * (1.0 + (rule_mode == 0 ? buffer_pct : -buffer_pct) / 100.0)
enter_long = ready and (rule_mode == 0 ? close > entry_level : close < entry_level)
exit_long = ready and (rule_mode == 0 ? close < anchor_mean : close >= anchor_mean)

plot(anchor_mean, "Anchored average", color=color.blue)
plot(entry_level, "Entry boundary", color=color.orange)
if exit_long and strategy.position_size > 0
    strategy.close("Long")
if enter_long and not exit_long and strategy.position_size == 0
    strategy.entry("Long", strategy.long)

For another view of volume, see On-Balance Volume. For the effect of a rolling window, see the simple moving average guide.