VWAP Indicator

The formula

For each bar take a price, often the average of the high, low and close, called the typical price, or just the close. Multiply it by the volume of that bar. Add up those values from the start of the session, and divide by the total volume so far.

A worked example

Take six made-up bars, with prices 100.0, 100.4, 100.9, 100.2, 99.8, 100.1 and volumes 500, 800, 300, 1,200, 1,500, 700. The total volume is 5,000 and the total value is 500600.00. VWAP is 500600.00 / 5,000 = 100.120. The simple average of the six prices is 100.233. VWAP is lower because the two heaviest bars, with volumes of 1,200 and 1,500, were at 100.2 and 99.8, below the lighter ones.

The two lines start together and then pull apart. After bar 3 they are close. After bar 5 the heavy volume at 99.8 has dragged VWAP down faster than the simple average.

VWAP as a benchmark for fills

Suppose a bot bought at 100.25 during that session. The final VWAP was 100.120, so the buy was 0.13 above it, which is 0.130 percent. A buy above VWAP paid more than the average trade did. Over many trades the gap shows whether the bot's fills are good or bad. It is a way to measure slippage against what the market did, not against what the bot expected. For a larger order the point is to avoid pushing the price, and the method is to slice the order over the session.

Why the weighting matters

A simple average treats a quiet bar and a busy bar the same. But most of what was actually traded happened in the busy bars, and those are the prices most people got. VWAP leans toward them. That is why it is used to judge fills: it is a fair guess at what an average participant paid. It also means a single large print can move it, and a thin market can give a VWAP that rests on a few trades. Check that the volume your data feed gives is real traded volume for the instrument. Some markets, such as spot currencies, have no single central volume, so a VWAP there rests on one venue or on tick counts and means less.

How a bot misuses it

• Mean reversion to VWAP. A rule that buys below VWAP and sells above it works in a range. In a trend the price can stay below or above for the whole session, and the bot keeps adding into it. See martingale.
• Early-session noise. In the first few bars the volume is small, so VWAP moves with each bar and means little.
• Different inputs. A close, a typical price and a tick price give different values, and some platforms count the pre-market. Check what yours uses before comparing.
• Costs. A rule that trades each cross pays the spread each time.

How to check a VWAP rule

Test it on separate days, trending and ranging, and count the trades. Use full costs. Keep data back and test once, as in backtesting. Check the max drawdown, which is where a rule that fights a trend shows its cost.

What it does not tell you

It does not say where the price goes next. VWAP is a description of what already traded, and the numbers on this page are made up. It is education, not a signal.