3-Cycle Prediction Engine?

A reader commented and asked: “Ned Davis Research calculates a time cycle composite. How good is an equal weighting of the annual seasonal cycle, the Presidential term cycle and the the decennial cycle at predicting the direction of the market?” One straightforward way to construct a forecast for a given month by equally weighting historical data at these three frequencies is to use an average of: (1) the average return for the calendar month up through the previous year (2) the average monthly return for the Presidential term year up through the previous Presidential term; and, (3) the average monthly return for the year of a decade up through the previous decade. Even assuming well-behaved distributions of monthly returns, such modeling requires very long sets of historical data (many decades). Using monthly returns for Dow Jones Industrial Average (DJIA), the S&P 500 Index for January 1950 through June 2011 and Shiller’s S&P Composite Index for January 1871 through June 2011, we find that: (more…)