The Behavior oJ Stock Market Aggregates: Evidence oJ Dependence on the American Stock Exchange* George C. Philippatos and David N. Nawroeki, The Pennsylvania State University The purpose of this paper is to investigate the short-term behavior of speculative prices on the American Stock Exchange (AMEX). The data consist of the daily proportions of securities advancing, declining, and remaining uncilanged in price; they are analyzed using some simple methods from information theory as well as by means of the more familiar serial-correlation analysis. The data also are adjusted for non-stationarity of sample means and for the effects of untraded securities upon the computed coefficients. It is concluded that, even after proper adjustments for known sources of bias, the estimated coefficients indicate the existence of considerable positive dependence of today's proportion on the AMEX to the respective proportion of yesterday. The best forecast for time t obtains by the assignment of a weight between .32 and .55 to the value in t-1 and the remaining weight to the long-run means of the respective proportions. Research employing similar methods has been carried on on data from a number of Stock Exchanges. For example, Theil and Leenders [14] examined the temporal behavior of the daily proportions of se- curities advancing, q~, declining, q2, and remaining unchanged in price, qs, on the Amsterdam Stock Exchange and concluded that t}ley were generated by a first-order Markov process; under this scheme the best prediction for tomorrow's proportion was obtained by assign- ing a weight of .49 to yesterday's observation, q,. t-l, and .51 to the long-run mean, ~. Dryden [2][31, applying similar methodology to data from the London Stock Exchange, concluded that the best pre- diction scheme for the London data was also a first-order Markov scheme which yielded optimal forecasts for t when the value at t-X was assigned a weight of .58 and the long-run mean a weight of .42. However, when the method was applied to data from the N.Y.S.E. Volume I. Number 2 Fall 107.7