TWAP (Time-Weighted Average Price)

The running average of price over time within a session — every bar counts equally, unlike VWAP where volume sets the weights.

What it is

TWAP (Time-Weighted Average Price) is the plain arithmetic average of price over a defined window, most often the current trading session: add up a price value (such as the bar's typical price) for every bar since the session began and divide by the number of bars. Its origin is execution benchmarking, not charting — institutional desks slice large orders evenly across time and measure their fills against TWAP to check execution quality. On a chart it appears as a smooth line anchored to the session open, similar in shape to VWAP but built without volume. The difference from VWAP is exactly one thing: TWAP weights every time interval equally, while VWAP weights each price by the volume traded there. On a low-activity bar, TWAP counts that bar's price fully; VWAP barely moves. Both are descriptive averages of what has already traded, not levels the market owes any respect to. This is educational material, not financial advice — no average line predicts future price, and trading involves risk of loss.

How it works

TWAP is a cumulative mean that resets at an anchor, usually the session or daily open. Each confirmed bar contributes one price observation — commonly the typical price (high + low + close) / 3, though close or hlc4 are also used. The formula is TWAP = (sum of price values since the anchor) / (number of bars since the anchor). Compare this with VWAP = (sum of price x volume) / (sum of volume) over the same window: structurally identical cumulative ratios, differing only in the weighting term. Because each new bar is one observation among an ever-growing count, TWAP becomes progressively harder to move as the session ages — a large price swing late in the day barely bends the line. Early in the session, with few bars accumulated, it tracks price closely. One honest subtlety: chart-based TWAP is an approximation of true TWAP, which is defined on continuous time; using one value per bar assumes each bar represents an equal slice of time, which breaks slightly around session gaps or missing bars. And because TWAP ignores volume entirely, it can sit far from VWAP on days where volume clustered at particular prices — that divergence is a description of how activity was distributed, not a signal.

How traders read it

• Price above the session TWAP means the current price is higher than the simple time-average of the session so far; below means lower — a descriptive statement about the session, not an instruction to act.
• Compare TWAP with VWAP on the same chart: when they diverge, volume was concentrated away from the time-average price (VWAP is pulled toward high-volume prices, TWAP is not).
• Expect the line to flatten as the session ages — with hundreds of accumulated bars, late-session moves shift TWAP only slightly, so a 'break' of TWAP late in the day reflects a stale average.
• As an execution benchmark: fills consistently better than the interval's TWAP suggest execution beat an even-paced schedule over that window — a cost measure, not an edge.
• TWAP behaves the same on instruments with unreliable or absent volume data (spot forex, some CFDs), which is precisely the situation where VWAP's weighting is least trustworthy.
• Remember the anchor defines the line: a TWAP anchored to a different session open or timezone is a different line, so state the anchor when comparing readings.

Common settings

The main choices are the price source (typical price hlc3 is common; close or ohlc4 also appear) and the anchor that resets the accumulation — usually the daily session open, though weekly or custom-session anchors are used. There are no length parameters to tune: TWAP is cumulative from its anchor by definition. Intraday timeframe affects granularity slightly (more bars approximate continuous time better), and execution desks typically compute TWAP over the specific interval an order works, rather than the whole session.

Strengths

• Fully deterministic and simple: a cumulative sum and a bar count, with no smoothing constants or lookback lengths to curve-fit.
• Works where VWAP cannot be trusted — spot forex and other decentralized markets lack true consolidated volume, and TWAP needs none.
• A recognized execution benchmark: measuring fills against TWAP is standard practice for evenly-paced order slicing.
• Does not repaint: each confirmed bar's contribution is fixed, and the line never revises its history.
• Clean disambiguation partner to VWAP — plotting both makes the effect of volume weighting directly visible on the chart.

Pitfalls to watch

• It is not VWAP: TWAP ignores where volume actually traded, so it can sit far from the volume-weighted 'average transaction price' that VWAP represents — using them interchangeably is an error.
• Cumulative staleness: late in the session the line is dominated by hours of old bars and responds sluggishly to new information; touches of TWAP at that point mean little.
• Anchor sensitivity: session boundaries, timezones, and DST shifts change where the average resets, silently producing different lines on different platforms for the same instrument.
• Bar-sampling approximation: one observation per bar assumes equal time slices, so gaps, halts, and missing bars distort chart TWAP relative to true continuous-time TWAP.
• Support/resistance folklore: there is no verified mechanism that makes price react to a TWAP line; treating it as a bounce level is an untested assumption.
• No predictive power: TWAP summarizes the session's past prices only — it forecasts nothing, and trading around it carries full risk of loss.

Pine v6 example

//@version=6
indicator("Session TWAP vs VWAP", overlay = true)

src = input.source(hlc3, "Price Source")

// Reset the cumulative time-average at each new daily session.
newDay = timeframe.change("D")

var float priceSum = 0.0
var int   barCount = 0

if newDay
    priceSum := 0.0
    barCount := 0

priceSum := priceSum + src
barCount := barCount + 1

twapLine = priceSum / barCount

plot(twapLine, "Session TWAP", color = color.blue)
plot(ta.vwap(src), "Session VWAP (compare)", color = color.orange)

Pro tip: Plot TWAP and VWAP together for a few sessions and study where they separate: the gap is a direct picture of how volume was distributed relative to time, which is more informative than either line alone. If your instrument has no trustworthy volume data (spot forex especially), be honest that VWAP there is built on broker-feed tick counts, and prefer TWAP for a defensible average. Educational context only, not financial advice — neither average predicts price, and trading involves risk of loss.

Educational only — not financial advice, not a recommendation to trade. No indicator is predictive; trading involves substantial risk of loss.