Volume or Dollar bars vs. volatility normalized and demeaned financial time series

Question

In his book - Advances in Financial Machine Learning, Marcos Lopez de Prado familiarises the reader with a number of ways of normalizing our financial time series data. Below I provide a couple of examples (source: here) for the purpose of this question:

1. Time Bars: Bars indexed by time intervals, minutely, daily, etc. OHLCV (Open, High, Low, Close, Volume) is standard.
2. Tick Bars: Bars indexed by orders, with each set # of orders (usually just 1) creating a distinct bar. Order price, size, and the exchange the order was executed on are common.
3. Volume Bars: Bars indexed by total volume, with each set # of shares traded creating a distinct bar. We can transform minute bars into an approximation for volume bars, but ideally we would use tick bars to maintain information for all parameters across bars.
4. Dollar Bars: Similar to volume bars, except measuring the total value in USD traded hands. An example would be $100,000 bars, with each bar containing as precisely as possible that dollar value.

When transforming time bars of prices to volume bars or dollar bars, the returns of the resulting time series are much closer to normally distributed.

Another very standard way of normalizing time serie would be to:

1. Volatility normalize per security
2. Demean the time series

My question is: Is there a significant difference between the normalization of Volume or Dollar bars vs. Vol normalize and Demean?

Answer 1

The dollar bars certainly allow for a partial recovery of normality through a price sampling process subordinated to a volume, tick, dollar clock.

It is well known that returns are assumed to be normally distributed but in reality they have a high kurtosis and fat tails, they are leptokurtic.

Dr de Prado posits in several papers but mainly in a well known paper titled The Volume Clock, that if we sample prices based on a volume clock rather than fixed time interval sampling, then we get better statistical properties.

The following is a jupyter notebook that shows the results on real data: Notebook.

In response to your question: Yes there is a significant difference between the techniques.