The New Market Wizards: Conversations with America's Top Traders

by Jack D. Schwager

In the coin toss example you just provided, computer simulations make it possible to generate huge data samples that allow you to conclude that the mean has no limit. But how can you definitively state that the variances of commodity price distributions are not finite? Isn’t the available data far too limited to draw such a conclusion? There are statistical problems in determining whether the variance of price change is infinite. In some ways, these difficulties are similar to the problems in ascertaining whether we’re experiencing global warming. There are suggestive indications that we are, but it is difficult to distinguish the recent rise in temperature from random variation. Getting enough data to assure that price change variance is infinite could take a century.