Detecting Information Structure And Measuring Information Risk In Chinese Security Market

Information is the fundamental basis for investors to develop investmentstrategies, which directly affects the price discovery process of the financial marketassets and asset price behavior. Based on a systemic perspective and considering thedifference of multi-dimensional information structure under different time scale, wedeeply studied the issue of detecting information structure and measuring informationrisk at low frequency and high frequency respectively. The summary of the maincontent are as follows:Detecting intraday information structure in Chinese stock market: At the intradayfrequency and based on the characteristics that the information doesn’t integrate intoasset price instantaneously, we applied Hidden Markov Model to model theunobservable state of stock information, and we built a transition probabilities matrixof information state to describe the dynamic association properties in temporaldimension. Empirical results prove that the model has effective ability to identifyinformation, and show that the information effects have characteristics of aggregationin Chinese stock market. The empirical research also estimated the information stateand strength of the samples. Further, based on the estimated transition probabilitiesmatrix of information state, the speed how fast information merged into informationwas surmised in Chinese stock market.The real-time measure of risk information of Chinese stock: Basing on thecharacteristics of time-varying intraday information composition structure, thenon-parametric computation expression of time-varying PIN was deduced, and amethod to measure the intraday high-frequency information risk was built. Applyingthis method, we designed an empirical research and proved that our model was able todetect the changing state of information risk and was also able to predict the suddenchange of asset price caused by the toxic information flow.Detecting inter-day information structure in Chinese stock market: At theinter-day frequency and based on the assumption that investors can infer the dailychanging composition structure of transaction types from the transaction data, aGARCH structure information model allowing the informed and uninformed arrivalrate to be time-varying and predictable was built. Applying the model we designed empirical research to investigate the dynamic process how Chinese stock marketinvestors learned from market transaction information and adjusted their tradebehavior. Basing on this, we constructed the inter-day time-varying probability ofinformed trade, we further studied the changing behavior model of informationstructure in real financial marketThe judgment of inter-day information structure and the measure method ofinformation risk in Chinese stock market: first of all, the symmetric order flow shocksitem was used to construct the information model in order to solve the problem thatthe numerical characteristics implicated in the former models cannot match the actualmarket order data. Secondly, based on the theoretical basis of the new model wesimulated arrival data of buyer initiated order and seller initiated order, and we drewits frequency distribution diagram, basing on which we judged what’s the informationstructure type of an individual stock and which model it suitable for. Finally, weanalyzed what’s the learning behavior of the investors theoretically and how thesebehaviors led to a stable information structure. Further, we expounded the principle ofinformation structure measuring in the state of low frequency, and we used the newmodel to measure the stock information risk.