A Complete Guide to the Futures Market, 2nd Edition

Appendix F

Practical Considerations in Applying Regression Analysis

I remember the rage I used to feel when a prediction went awry. I could have shouted at the subjects of my experiments, “Behave, damn you, behave as you ought!” Eventually I realized that the subjects were always right. It was I who was wrong. I had made a bad prediction.

—Burrhus Frederic Skinner

■ Determining the Dependent Variable

The title of this section might sound trivial. After all, the dependent variable is what we wish to forecast. However, in a price-forecasting equation, the selection of a dependent variable is by no means obvious. The following choices must be made:

1. Should the price be stated in nominal or deflated terms?
2. Should the price be based on cash or futures?
3. If the price is based on futures, should it be based on a nearest futures price series or a single contract?
4. Should the price represent the entire season or only a specified segment of the season?

The answer to question 1 would typically be deflated prices, unless a trend variable is included in the equation, in which case nominal prices may be a better choice. If, however, the equation does not include a trend variable, then the use of nominal prices implicitly assumes that equivalent fundamental conditions in two widely spaced years should result in approximately equal price levels. Obviously, this assumption is wrong. All else being equal, inflation will result in considerably higher prices in the more recent year. The subject ...