The first thing one should know about serial dependenceis that it has nothing to do with an addiction to Rice Krispies, cornflakes or oatmeal. Serial dependence refers to the notion that returns evolve nonrandomly; that is, they are correlated with their prior values. One variation of serial dependence is called mean reversion. With mean reversion, returns revert to an average value or asset prices revert to an equilibrium value. If an asset is priced above its equilibrium value, its price will not change randomly; it will be more inclined to decrease than to increase. Conversely, if an asset is priced below its equilibrium value, it will be more likely to increase than to depreciate further. Another variation of serial dependence is known as trending. In a trending pattern, a positive return is more likely to be followed by another positive return than a reversal, and a negative return is more likely to be succeeded by another negative return than a positive return. Of course, some returns may conform to nonrandom patterns that are more complex than simple mean reversion or trending. For example, the returns in a series may be correlated not with their immediately prior returns, but with more distant prior returns. Alternatively, returns may be linearly independent of prior values but display serial dependence after some transformation. The extent to which asset returns evolve nonrandomly has important consequences for financial analysis. First of all, if asset returns are nonrandom, then their variance will depend on the interval used to measure them. Instead of varying proportionately with the time interval, the variance of returns will vary at a varying rate. I will discuss some of the implications of this nonlinearity later. Second, if investment returns are serially dependent, they are at least partly predictable. This result is of obvious interest because it raises the possibility that we can devise trading rules to generate abnormal profits.