Good asset allocation is all about having the right weights placed in the right asset classes. The less positively correlated the asset classes, the better the diversification benefit. Having a well-diversified portfolio allows you to have a higher portfolio return for a given level of volatility, or equivalently, lower volatility for a given portfolio return.

Fund managers re-balance a portfolio periodically by cutting down on an asset class whose allocation becomes larger than expected and/or beefing up an asset class whose allocation has become smaller than expected. The rationale for this is because asset class returns are expected to always return to their means. And the expected portfolio return is a weighted average of the mean returns of the asset classes.

The mean is after all an average --- a statistical measure of central tendency. So asset class returns should always return to their means. Or do they?

In general, asset class returns do return to their means except they may not be the means that were originally calculated! This is because return distributions are not perfectly normal in the statistical sense of the word. And diversification calculations are hinged on normal distributions. If returns were indeed normally distributed, they would have an equal probability of being above and below the mean.

A

*greater-than-normal*number of extreme values in the tail results in what is called a “fat tail” and this skews the distribution so that probabilities are no longer equal above and below the mean. Non-normality can be intrinsic or it can also be the result of a non-stationary series. A non-stationary series simply implies that the mean is not constant over time.

For the uninitiated, these concepts can be a little abstract at first glance. For now, simply realize that the mean (also called the expected value in statistical parlance) is for the most part a

*moving target.*