Spreads, Betas and Risk

In the previous chapter, we concluded that no noticeable autocorrelation exists for daily returns. This implies that knowing the level of the previous day’s return does not help us in forecasting today’s return. The hypothesis we will formulate in this section is that we can artificially create a time series that is somewhat forecastable. We will refer to this new time series as a spread. The claim we are making is that a stock spread has a better chance of being tradable than an individual outright does. This, of course, is a subjective and suspicious statement in and of itself. Bear with me, though for the remainder of the chapter. I am mostly interested in conveying a methodology of thinking rather than a concrete fact about price behavior.