Economic foundations of technical analysis
Date of Issue2013
School of Humanities and Social Sciences
Though technical analysis has gained huge popularity among the practitioners for over two centuries, it is still known as "voodoo finance" to academicians due to lack of solid theoretical supports. Among all the technical indicators, price patterns and volume signals attract the most disputes due to their subjective identification processes. While, many pieces of empirical evidence have been found to support the profitability of technical analysis, showing that those indicators indeed help the speculators to "beat" the market, no one theoretical or statistical model has been developed to replicate all price movements and stylized facts that are documented in financial market. Besides, no existing literature can further justify how the price trends following the chart patterns are pre-determined. Therefore, the aim of this thesis is to go a step further to develop a solid theoretical model to replicate those charting indicators, specifically the price patterns and the visual price volume relations. I approach this goal from the standpoint of technicians that "Price are determined by the demand and supply" and "History repeats itself", by extending the classic work of Day and Huang (1990). Such extension leads to a new theoretical framework where both price and volume series are simultaneously determined by the endogenous buying and selling orders from two types of market players, fundamentalists and chartists. In the thesis, I show that (1) the seemingly chaotic fluctuations in price and price volume relations in the real market can be simulated with high compatibility; (2) most of the commonly-seen chart patterns and their following predictive powers can be easily replicated, (3) most of the price volume relations found in researches, not limited to the volume signals in chart patterns, can be simulated, (4) popular stylized facts documented by extensive literature can be captured, and (5) the nonlinear causality relations between the simulated price volume series is confirmed using the nonlinear Granger causality test as well. This thesis also goes further to provide economic arguments to the rationale of technical analysis. It is demonstrated that supporting zones and resisting zones can be developed and once price falls into these zones, the previous trend will be reversed and a reversal point occurs. As a result, chart patterns can be used to predict the trends and help speculators to detect these supporting zones and resisting zones. Plausible economic justifications can also be offered to the relations between the asset returns and volume. Besides the theoretical supports to the visual technical indicators, I also empirically examine the volume signals in chart patterns. Perceptually Important Point (PIP) Identification Process is used to detect different patterns in three Asian financial markets, namely the Hong Kong Hang Seng Index, the Singapore Straits Time Index and the Japan Nekkei 225. By comparing the conditional returns on the chart patterns to the original unconditional returns using goodness-of- t test, I con rm that most of the chart patterns are informative in the three Asian financial markets except the double tops and double bottoms. Furthermore, our study distinguishes from previous empirical literature by introducing the corresponding volume signals to the evaluations, and it is shown that the previous unsatisfactory results of double tops and double bottoms in chi-square test and most of the results in Kolmogorov-Smirnov test can be greatly improved.
DRNTU::Social sciences::Economic theory