Direction Indicators in Financial Modelling

Abstract

This article adds to the debate in the literature on both long memory processes and technical analysis in financial modelling. Recently, results have noted the apparent long memory property powers of absolute returns in high frequency asset returns data. This has led to the formulation of long memory time dependent conditional heteroskedastic processes such as FIGARCH and corresponding long memory stochastic volatility processes. The long memory volatility processes appear to be superior to other parameterisations.

However, the processes are incomplete. Limitations are in the lack of a directional indicator and the incomplete use of all available price information. Such inefficiencies are discussed here as alternatives to the Wiener-Kolmogorov prediction theory, and the usefulness of Japanese candlesticks. To complete this task the superior results of asset returns have to be re-interpreted in terms of asset prices.