Ready Reckoner for Index Pairs

As you have seen over the course of this week, one of the keys to working on this strategy is to find well correlated pairs. You have seen that things like the Australian banks and RIO and BHP are good candidates because of the homogenous nature of their business. Given the nature of the GICS sectors in the Australian sharemarket you need to be careful when it comes to correlations. Just because 2 companies are in the same sector has little to do with how correlated they will be.

What I have prepared for you today are some correlation matrices for the major index futures around the world – which is a very good proxy for index CFDs. While correlations change over time it’s important to know at least the ones that have a good history.

I have used the underlying index name in the table here but keep in mind this is the correlation between futures and not the index itself. This is the most relevant measure for traders in this particular case.

You can now see that by looking for pairs to trade you need to isolate those that have a good correlation between one another. Unfortunately though the highest correlations (like the Dow and the S&P) may not be ideal because there is unlikely to be huge spreads occurring like you would see between other indices.

As you can see here – a spread has existed between the Dow and the S&P – however you can see that they have run largely parallel for much of the observation period – which is less than ideal for this type of trade. You can see alternatively that the Aussie has initially spread away but has then crossed over the value of the US indices – this is much more what you are looking for. So you need lots of correlation….but not too much.

This entry was posted in Uncategorized. Bookmark the permalink.

1 Response to Ready Reckoner for Index Pairs

  1. Du says:

    If the Dow and the SP runs in parallel with such a high correlation, in my point of view, won’t it be actually better to adopt this strategy but at a shorter time interval? High correlation provided solid risk shield so whenever a high random fluctuation we are more certain there would be a convergence. A lepotkurtosis shape would even be preferred.

Comments are closed.