City Research Online

Abstract

The aim of the thesis is to evaluate the ability of technical analysis to predict movements in financial markets. This study is of obvious practical interest in that technical analysis is used intensively by market practitioners. It is useful to know whether there is any objective evidence that it works. The study has also proved timely in that technical analysis has stimulated a small but insightful programme of academic research in the past decade, and our work adds to that line of research. This study differs from previous academic research in one important way. All of our empirical work derives from information on the forecasts and trading recommendations of analysts themselves. This contrasts with most earlier studies, which try to mimic the forecasts of technical analysts by applying mechanical trading rules. Specifically, we utilise data from (a) a specially conducted survey of a group of analysts through 1998, and (b) unique data sets on daily published forecasts and trading recommendations by a leading provider of technical commentary through the years 2000-2001. Our analysis of this data strongly suggests that technical analysis does have value, and that the behaviour of technical analysts cannot be modelled using simple (or even quite complex) mechanical trading rules.

A widely accepted definition is that “Technical Analysis is the study of market action, primarily through the use of charts, for the purpose of forecasting future price trends” (J.J.Murphy, 1986). In contrast to “Fundamental Analysis”, technical buying and selling strategies are based on the observation of past history activities, extracting market psychology from price patterns. The topic is important because technical analysis is by far the most common method used for short term forecasting by traders in financial markets. In spite of this, until recently there has been very little serious academic work on the value of technical analysis.

The prime reason for the paucity of academic work is the fact that technical analysis does not involve well-defined statistical procedures. Rather, technical analysis is an umbrella term for a very diverse collection of techniques, some quantitative and some judgmental, most with little scientific basis, and often sold with exaggerated claims about their likely success. This has made technical analysis an easy target for ridicule, most notably in Burton Malkiefs classic Random Walk down Wall Street (Malkiel, 1973).

In recent years, several factors have caused researchers to take technical analysis more seriously. From a number of “forecasting competitions” it is now well understood that a catholic approach helps improve forecast accuracy in a wide variety of business applications. The poor performance of many popular linear time series forecasting methods, notably Box-Jenkins analysis, in these competitions has shown that overselling is not unique to the world of technical analysis. Nonlinear modeling and forecasting methods are now in vogue. Indeed, in the financial markets, it is now recognised that many different regimes can be at work across a single time series of market prices, so it looks much more reasonable to use a set of tools rather than search for a single underlying model.

The availability of time series data based on high-frequency financial market prices has made it easier to make objective assessments of competing forecasting methods, and a number of academic studies have exploited this to evaluate technical analysis. However, this work has focussed on the narrow area of easily-replicated mechanical trading rules, such as moving averages and filters (for example, Brock et. al. 1992), and a few well-defined turning point patterns, such as the “Head-and-Shoulders” (Osier and Chang, 1995). Just as it would be hard to argue that a single-equation regression analysis would provide a good test of the value of conventional econometric forecasting, we argue that this focus on quantitative rules does not adequately reflect the complexity of the way technical analysis is applied in practice.

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HG Finance
Departments:	Bayes Business School > Bayes Business School Doctoral Theses Bayes Business School > Finance Doctoral Theses

Export

Downloads