Investors' Behavior Based On High-frequency Data

With the rapid development of the securities market, also in the rapid progress ofinformation technology and data resolution, the financial data from the original monthly,weekly and daily data gradually to fine to hours, minutes or even seconds of data, the ultrahigh frequency data acquisition in our financial field, especially in the study of behavioralfinance in the era, become of extremely important.Research of behavioral finance combines with psychology and finance, it’s think thedecisions of investors is not only restricted by the objective market, but also by itself oraround groups of subjective influence, so the investors will show various abnormal, andcorresponding reflected for a variety of vision in the stock market, for the interpretation ofthe vision also emerge in an endless stream.In this paper, focusing the behavior of investors, starting from the behavioral financetheory, applying high frequency data for (ultra) Autoregressive Conditional DurationModel(ACD model), through the reasonable extension for its basic form, get the differentforms of the corresponding. Aiming at the market microstructure, impact of event andinvestor sentiment aspects, not only do the theoretical analysis, and do the empiricalanalysis using the Ping An Bank stock transaction data from the combination of theShanghai and Shenzhen300stock samples.Firstly, by constructing the price duration,introducing the three representative variables of market microstructure--bid ask spreads,each average trading volume and density in the basic ACD model, respectively and jointlyanalyzes the relationship between duration and them. The results found, three and durationhave significant negative correlation, and explain this phenomenon based on the microstructure theory, think this is due to the existence of information traders, rather than causedby mobility.Secondly, through constructing volume duration, considering AsymmetricEffect in the basic ACD Model, analysis the impacts of events attack on investor behavior.Found in the expected impact of event, asymmetric effect is not big, but in the unexpectedevent (for example, China Everbright events in August16,2013) impact, asymmetric effectis larger, accordingly obtained the reaction process of investor behavior before and afterthe incident shock.Finally, through the analysis of the measure of investor sentiment index,choose the exchange rate as a representative variable of investor emotion using case bycase data, and the direction of buy and sale respectively on behalf of optimism and pessimism. Empirical findings, both the rate of return, or volatility, are positivelycorrelated with investor sentiment.Through these research findings, and then put forward some suggestions, such asinvestors and the government through the observation for duration, in the face of durationis too large or too small, namely risk larger or smaller, should reduce the transaction, andtake measures to avoid the risk. In addition, in order to solve the problem of marketliquidity, reduce or even eliminate the noise impact of information transactions, the marketregulators should focus on improving the market mechanism, make the information moretransparent, effectively solve the problem of information asymmetry, only in this way, canthe better and faster development of China’s securities market.