Review of statistical modeling in technical analysis of financial markets

Abstract

Predictability of asset returns in share market has been an immense interest over the past decades and Statistical Modeling has been playing a vital role in it. This paper reviews statistical modeling in Technical Analysis of financial markets. Linear and non linear regression models, Vector Auto Regression (VAR) Models and Spectral analysis found tested on share return and trading volume. Some common weaknesses were identified in reviewed articles. Authors have not reported the results of modeling assumptions, independence, normality and homoscedasticity of errors. Model verification criteria and results also not reported. Hence findings of their studies were not reliable. Majority of studies were focused on developed markets and very few attempts on emerging markets. Only two studies were found in Sri Lankan context and their results were contradictory. It is recommended to test GARCH /ARCH models and Spectral Analysis in Sri Lankan context.