In one of the previous posts, we introduced the Alpholio™ app for Android. This post is the fourth one in a series covering the app’s services in more detail.
The Correlation service produces charts of the correlation coefficient between returns of two securities over a specified period. The correlation coefficient ranges from -1 (perfect negative correlation; returns always moving in opposite directions) to +1 (perfect positive correlation; returns always moving in same direction). A correlation coefficient of 0 indicates that returns of the two securities are unrelated or random with respect to each other. A correlation coefficient of close to +1 does not imply that the two securities are virtually identical (see our detailed explanation).
Knowing which securities are highly correlated and which are not is useful in portfolio construction. Generally, a security with a correlation coefficient of 0.6 or less in relation to others is considered a good candidate for portfolio diversification. Variations in the correlation coefficient over time are also important, especially in periods when one of the analyzed securities severely underperforms (see below).
To access the service, start the app, open the navigation drawer and tap the Correlation item:
This will open a new screen, on which you can enter inputs for the chart. To expand the Return Frequency and Span sections, simply tap each section header:
To change either ticker, tap the corresponding field and use the pop-up keyboard to edit it. (If you need to find the ticker based on other information, use the Security Lookup service of the app.)
To modify either the From or To date, tap its corresponding button. This will pop up a standard date selection dialog. The From date must chronologically precede the To date.
To select a different return frequency, tap the corresponding radio button. Generally, monthly returns will provide a smoother correlation plot than weekly or daily ones.
To change the rolling correlation window, tap the Span field and use the pop-up keyboard to edit it. The span is expressed in the same units as the return frequency and has to be a whole number greater than one.
After you specify all parameters, tap the Get Correlation button. If any of your inputs are invalid, you will see a brief pop-up warning. If all settings are acceptable, they will be saved on the device for subsequent use. Please note that to generate the chart, your device must be connected to the Internet.
After the app obtains and processes the data, you should see the following screen:
The first thing you may notice is that the chart begins in October 2005 and not January 2000 that was specified as the From date. That is because the inception date of AGG was in September 2003, and two years worth of monthly returns were required to calculate the first data point. The app automatically selected the largest possible date range for the analysis.
To zoom in on a portion of the chart, tap the + button or use a spread gesture. To scroll a zoomed-in chart horizontally or vertically, use a corresponding swipe gesture. To zoom out, tap the – button or use a pinch gesture. To immediately restore the chart to its original view, tap the 1:1 button.
Below the chart, there is a Statistics section that can be collapsed and expanded by tapping its header. Here you can see that VTI and AGG returns were largely uncorrelated (both the mean and median correlation coefficient is very close to zero). However, at times this was not true, as indicated by the minimum and maximum values.
The forecast field projects the best estimate of correlation in the next time increment (in this example, October 2014). The estimate is calculated in a dynamic manner, using the entire data set of the chart.
In this example, you can see that the return correlation between the broad US equity and bond ETFs dramatically increased at the onset of the financial crisis in 2008 and did not subside until about three years later. One possible explanation: AGG is composed of Treasury, MBS and corporate investment-grade bonds, of which the last ones underperformed in 2008, similarly to their stock counterparts. After the equity market rebounded in early 2009, low interest rates caused a sustained positive correlation of bond returns with those of stocks.