Identification of market forces in the financial system adaptation framework

Abstract

The behavior of financial markets is modeled by a feedback adaptive system. Based on the internal model identification results, a time-varying state space model with instrumental variables is proposed as the adaptive filter to introduce external influential factors of the stock market to the system. One-step-ahead prediction results are obtained through the optimization of hyperparameters followed by the Kalman filter estimation process. The interest rate, oil price, Baltic Dry Index, CBOE DJIA Volatility Index and exchange rate of Euro to Japanese Yuen are tested in this paper as the economic and sentiment indicators. Testing results suggest that the determinant indicators of the stock market vary in different periods. Combining some of them can explain a significant proportion of the market volatility. By these results, our framework is also evidenced to be effective with the correct input. © 2010 IEEE.