Jul 08 2009
Market Correlations, And Analytic Errors
Kathy Lien posted on the subject of stock/forex correlations on Tuesday. She focused on the relationship between the S&P 500 and EUR/USD and USD/JPY, talking about how those linkages have changed over the course of the year. Some of the shifts have been relatively minor, but some have been significant. As I pointed out in my Traders Expo webinar, watching the changing links between markets can tell you a lot about what’s going on in the big picture. Correlations are also a subject I spent some time on in my book, The Essentials of Trading.
Here’s where I have to take Kathy to task a little bit, though, and potentially educate my readers a little bit too.
In her post Kathy said the following:
In comparison, since the beginning of the year, the correlation between these two instruments has been greater than 80 percent. This meant that 80 percent of the time that stocks rallied, the EUR/USD strengthened as well.
Tsk, tsk Kathy.
As any statistics student will tell you, correlation does not mean x% of the time this happens when that happens. In simple terms correlation means the amount of variance in the dependent variable (the forex pair in Kathy’s example) which can be explained by variance in the independent variable (stocks).
The reason this makes a difference is that the 80% thing could simply be reflecting the fact that the big moves in forex rates coincide with big moves in stock prices. In other words, stocks and forex could be going their own way on most days when the price changes are minor, but trade in close tandem on those dates when the markets are strongly directional one way or the others. This can be an important distiction, especially for a short-term trader.
Here are some other posts which might interest you:


