Common-Stock Transaction Sequences and the Random-Walk Model

This paper reviews the current status of the random-walk model as it applies to price-change sequences in common stocks, and examines certain striking dependencies in these sequences at the transaction level. In particular, successive transaction-price changes are seen to be negatively correlated, with each price change projecting its negative influence forward toward its successors in the sequence. The strength of this influence is diminished sharply by each nonzero price change that it is required to “penetrate,” but is only slightly weakened when projected across changes of zero magnitude. In the latter case, in fact, statistical dependency is shown to exist across intervals of as many as ten transactions or more.