Asset pricing and portfolio choice with technical analysis

Abstract

Technical analysis is the study of market movements, primarily through the use of past prices and volumes, for the purpose of forecasting future price trends. Despite its popularity among practitioners, academics tend to be skeptical about its true usefulness. One of the major reasons is that it lacks a theoretical basis in finance theory. Although there is increasing empirical evidence in favor of its effectiveness, the empirical debate remains unsettled, meanwhile the progress on strengthening its theoretical basis is relatively slow. To understand better technical analysis as an important and popular investment tool, this thesis aims to further tie technical analysis to modern finance theory in an attempt to tighten this gap in the literature. This thesis includes two chapters that study portfolio choice problems and two additional chapters that study asset pricing problems, in which investors make strategic use of information from technical analysis, specifically the moving averages. Our model approach provides several new insights to the field. We develop a model to examine the effects of the uncertain predictive power of moving averages on portfolio choice. We find that investors accounting for such uncertainty allocate substantially less wealth to stocks and are more conservative in market timing for longer horizons. Furthermore, the utility loss of ignoring this uncertainty can be sizable and increases with horizon at an increasing rate. We present another portfolio choice model to theoretically illustrate that moving averages can be useful for investment when stock returns are correlated. We also formulate an asset pricing model and propose some plausible equilibria in which future prices can be predicted by moving averages. This model provides a theoretical basis for some recent empirical findings that moving averages have predictive power. We further formulate a similar asset pricing model which emphasizes development of estimation and testing strategies to empirically test the proposed equilibria. Using S&P 500 index and dividend data for the period January 1871 to December 2015, we empirically reject the possibility that investors’ trend following behaviour is the driver of the stock market in the long run.