The GARCH Learning Behavior of Investors' Forecasting Transaction Arrival Rate of Chinese Stock Market

Based on the assumption that investors have the learning ability,a GARCH structure information model allowing the informed and uninformed arrival rate to be time-varying and predictable was built. Applying the model and basing on the Shanghai & Shenzhen 300 Index component stocks from 2006 to 2009 high-frequency tick by tick transaction data, we investigated the dynamic process how Chinese stock market investors learned from market transaction information and adjusted their trade behavior. We found that investors could optimize their next period transaction decision-making based on previous forecasting arrival rate and current period arrival rate derived from order data. In addition, the impact of prediction error on next period forecast is a decreasing function whose second derivative is greater than zero.