Detecting Turning Points of Trading Price and Return Volatility for Market

Trading agent concept is very useful for tradingstrategy design and market mechanism design. In thispaper, we introduce the use of trading agent for marketsurveillance. Market surveillance agents can bedeveloped for market surveillance officers andmanagement teams to present them alerts and indicatorsof abnormal market movements. In particular, weinvestigate the strategies for market surveillance agentsto detect the impact of company announcements onmarket movements. This paper examines the performanceof segmentation on the time series of trading price andreturn volatility, respectively. The purpose ofsegmentation is to detect the turning points of marketmovements caused by announcements, which are useful toidentify the indicators of insider trading. Theexperimental results indicate that the segmentation on thetime series of return volatility outperforms that on thetime series of trading price. It is easier to detect theturning points of return volatility than the turning pointsof trading price. The results will be used to code marketsurveillance agents for them to monitor abnormal marketmovements before the disclosure of market sensitiveannouncements. In this way, the market surveillanceagents can assist market surveillance officers withindicators and alerts.