Today’s post is a follow up of my previous post Looking at Correlation. In the first post I looked at correlation of the 7 ETFs that are used in the WTAA strategy that is one of the featured strategies of this blog. I provided Amibroker code to create a correlation table and code to create a correlation indicator.

In this post I will be exploring the correlation between the 3 strategies described in the blog: WTAA, MEOM and DVI strategy.

I am using Amibroker as my preferred platform for analysis. To look at the correlation we first need to create the equity curves of the individual strategies. This can be achieved by running one strategy and then rename the ~~~Equity ticker (this ticker represents the strategy equity curve after backtest and can be found in Group253) to a convenient name. This procedure can also be automated by using some code in the Custom Backtest File. Send me an email if you like to receive a copy.

After having created three equity tickers, we need to add them to a watchlist (in my case I use WL 10 for this). Now we can start using the equity curves as ordinary tickers at which we can apply any indicator or explorer.

First let’s create the correlation tables for 1 month and 1 year. What we see is that the longer term correlation is pretty low (good!) for MEOM. This is logical as the strategy trades only few days per month. However it provides a good diversification. I am also satisfied with the long term correlation between WTAA and DVI strategy. For me, 0.41 is still acceptable.

Let’s look at the correlation graphs to see how volatile the correlation of strategies is. The top graph is 21 days (1M) correlation and the bottom graph is 252 days (1Y) correlation. What I like about the long term correlation is that is remains pretty stable. The picture below only covers the period 2008-today, however the period 2003-2008 shows similar results in my tests.

It will be interesting to see the results of a portfolio trading these 3 strategies. My next post will cover this topic.

Happy trading!

QD

Hi

“…then rename the ~~~Equity”

I renamed it in group253 but if I want add it to watchlist from that group I have only one ~~~EQUITY- the latest ???

excellent. nice blog

When looking at correlations, consider spliting tables based on equity curve status. For example, if equity curve of System A is above its 50 day EMA, what is its correlation with others. Similarly, what is correlation when below. Another split could be based on volatility of equity curve. The objective is to identify periods of negative semi-correlation, and avoid being long during those periods.