A state space modeling for proactive management in equity investment

Abstract

This paper proposes a novel state-space approach to explain stock market dynamics driven by different types of trading, which leads to a new promising scheme for proactive risk management in financial investment. Particularly, it is assumed that the current price changes are formulated through daily trading by multiple types of traders, each of whom follows a specific investment strategy based on technical indicators and a fuzzy logic using past data of stock prices, volumes and yield curves. Moreover, the current price changes are represented by a linear combination of those multiple trading types, where the coefficients corresponding with the size of impact on the price changes are regarded as time-varying state variables to be sequentially estimated under a state-space framework. Thereby, this work develops a new factor decomposition method on price changes from a perspective of different traders’ demand and supply to analyze the current situations and potential risks in financial markets. In empirical experiments, it is shown that the implementation of particle filtering algorithm makes it possible to replicate market price changes. Further, new signals based on the estimated states are developed, which are applied to proactive risk management in financial investment. Especially, it has been found that the demands of yield curve-based traders subtracting those of trend-followers could be a promising signal of stock market crashes, which has successfully enhanced simple buy-and-hold strategy of SP, as well as constant proportion strategies.