Volume-Weighted Average Price (VWAP)

Formula

VWAP = Σᵢ pᵢ·vᵢ / Σᵢ vᵢ

pᵢ : price of trade i        (bar data: typical price (H + L + C) / 3)
vᵢ : volume of trade i
Σ  : over all trades in the horizon (usually one session, reset at the open)

Execution cost vs. VWAP (bps) = side × (P̄_fill − VWAP) / VWAP × 10⁴
side = +1 for buys, −1 for sells; positive = worse than benchmark

Institutional usage

• Execution benchmark. Brokers and buy-side trading desks report slippage against interval VWAP. Beating VWAP on a buy means the average fill came in below it.
• VWAP algorithms. These split a parent order across the day in proportion to a forecast intraday volume profile, so the fills track the market's own VWAP. The main risk is a bad volume forecast.
• Intraday signal. Price relative to running VWAP is a common mean-reversion and trend filter. Anchored VWAP measures from a chosen event, such as earnings or a gap.

Pitfalls

• Bar-based VWAP built from typical price only approximates trade-level VWAP. Use tick data when the result is used for cost attribution.
• Decide whether opening and closing auctions are included. On A-share markets the 09:15–09:25 call auction and the 14:57–15:00 closing auction can carry significant volume.
• A large order is part of the VWAP it is measured against. Participation above roughly 10–20% of volume makes the benchmark easy to "beat" and less informative; compare with arrival price as well.

Python reference implementation

Requires numpy, pandas. Tested before publication; the generator re-executes it on every rebuild. MIT licensed. Download vwap.py · All snippets

import numpy as np
import pandas as pd

def vwap(price: pd.Series, volume: pd.Series) -> float:
    """Session VWAP: sum(p * v) / sum(v)."""
    v = volume.astype(float)
    return float((price * v).sum() / v.sum())

def intraday_vwap(bars: pd.DataFrame) -> pd.Series:
    """Running VWAP that resets each trading day.
    bars: DatetimeIndex, columns high, low, close, volume."""
    tp = (bars["high"] + bars["low"] + bars["close"]) / 3.0  # typical price
    pv = tp * bars["volume"]
    day = bars.index.normalize()
    return pv.groupby(day).cumsum() / bars["volume"].groupby(day).cumsum()

def slippage_bps(avg_fill: float, bench_vwap: float, side: int) -> float:
    """Execution cost vs VWAP in basis points. side=+1 buy, -1 sell.
    Positive = worse than VWAP."""
    return side * (avg_fill - bench_vwap) / bench_vwap * 1e4

FAQ

Is VWAP the same as TWAP?

No. TWAP weights every time interval equally; VWAP weights by traded volume, so it follows the market's intraday volume curve.