Using VWAP to Estimate Daily Execution Costs in Backtests

Summary

The document asks how to estimate execution-price deviations and slippage for stock strategies when only daily OHLCV data is available. The proposed starting point in the question relates the daily high-low range to trading volume, with a restriction on order size relative to daily volume. The response recommends using volume-weighted average prices (VWAP) in backtests to make execution assumptions more transparent and reproducible.

The answer notes that broker accumulation algorithms may target performance within a range of VWAP, which makes it a practical benchmark for execution modeling. It does not provide a formula for converting order size into expected slippage, compare VWAP with the suggested high-low measure, or quantify the benchmark range. Accordingly, VWAP offers a useful backtesting reference but does not, by itself, establish likely fills or market impact for a particular order. The discussion is brief and does not include supporting data or literature references.

Key ideas

• VWAP can serve as a more transparent execution benchmark in backtests.
• The response does not specify a daily OHLCV formula for estimating slippage.
• Broker accumulation algorithms may be evaluated against a VWAP performance range.
• A VWAP benchmark alone does not quantify the market impact of a particular order.

Tags

# Estimate price movement per unit of volume for daily data


# Estimate price movement per unit of volume for daily data












I'm working on backtesting a number of stock trading strategies and need to estimate how much the execution price will likely deviate from the historical close price for that asset using daily data; which would be used to calculate an optimal position size and estimate probable slippage. Ideal would be a rough equation to get started with, that would justify spending the time/money on historical market depth data and more extensive research.

I intend to use Market on Close orders where possible, but would appreciate any rules of thumb, experiences or literature references that any of you may have.

E.g. would an equation like price_change_per_volume = (log(high) - log(low))/volume where volume is not more than 5% of total volume traded that day (as suggested elsewhere) be worse than nothing? How can I do better using only OHLCV daily data?

## Answer by Brian B (score 2)

https://quant.stackexchange.com/a/3790

You'll have a more transparent and reproducible result if you use volume-weighted average prices (VWAP) in your backtesting instead. Many brokers guarantee performance of their accumulation algorithms within a certain range of VWAP.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.