Often two (or more) items might have a mutual relationship with each other. No we aren't discussing personal, professional, romantic or platonic relationships but rather one of mathematical and statistical nature which can be employed when trading or investing.
Take the clothing industry, for example. We often see that the behavior of seasonal clothing is dependent upon the season we approach. As the frost settles on our lawns, we see an increase in sales of winter coats and cozy mittens. As it becomes colder, demand for winter wear skyrockets, yet as sunny days approach, we begin to find ourselves more intrigued by the latest trends of spring season. If we were to plot these trends in dropping temperature against sales of winterwear, we would see an indirect relationship; as temperatures dip, sales increase but as temperatures rise, sales decrease. If we were to plot the trends of dropping temperatures against the sales of summerwear, conversely, we would see a direct relationship; as temperatures dip, sales dip and as temperatures rise, sales rise.
How can we describe this?
Direct Relationship -> Positive Correlation --> The two variables move in unison
Indirect Relationship -> Negative Correlation --> The two variables move in opposite directions
Can this correlation be measured mathematically? Absolutely! If we have the data points for the pair at different points in time, we can use the statistical formula:
The above formula is for the Pearson Coefficient Formula. We will discuss the formula further in depth in later posts, but for now it's important to understand the importance of formula yields. The formula can yield a number between 1 and -1. For the results above 0.3 through 1, there exists a positive correlation between the two items tested; that is, the two items move in unison or rise/fall together. The higher the number, the more positive and better the correlation. For results below -0.3 through -1, there is a negative correlation between the two items tested; that is as one item rises, the other falls or vice versa. The lower the number, the more negative the correlation. Results between -.3 and .3 generally do not matter to us traders as there is often not a strong enough statistical relationship between the items in that range.
If you use tools like Microsoft Excel, it is pretty easy to calculate correlation. In the image, correlations are calculated between the four major index ETFs:
So, how does this apply to trading, again? If we know that there is a high positive correlation between two underlyings and if we see a divergence in this correlation, we can assume that the historically positive correlation between the two has broken and we can place a trade that reverts back to the mean. This we will cover in detail when discussing "pair's trading".
While correlation is a great tool to place trades, not all correlations are tradable as numbers alone tell the whole story. Here's a link to site that lists some funny or spurious. Who knew that per capita consumption of chicken correlates pretty highly with total US crude oil imports!
Eighteen-year old trader, future connoisseur of options.
Follow me on Twitter!