Modification Of VPIN Model And Prediction Of Volatility

Looking back at the historical development of financial markets in major economies,we can find that the abnormal fluctuation of asset prices is not rare in history,but it does great harm.The abnormal fluctuation of independent asset prices often leads to panic selling of investors,thus forming herding effect,which accelerates the deviation of asset prices from the reasonable value center level,thereby endangering the development of real economy.For investors,especially those with high leverage margin trading,unfavorable abnormal fluctuations may lead to investors'liquidation and large unexpected losses.In addition,abnormal volatility itself plays a self-reinforcing role.When abnormal volatility occurs,financial assets are likely to devolve the influence of investors' herding effect and magnetic attraction effect to the original trend movement,causing irreversible damage to the market itself.Therefore,to sum up,it is necessary for us to give full risk warning to potential abnormal fluctuations,and VPIN model will play an important role in the prevention and supervision of abnormal fluctuations because of its concise and accurate characteristics.Starting from the VPIN model proposed by Easley D,O'Hara(2011a),this paper focuses on the modification of VPIN model and its application in volatility prediction.In the revision of VPIN model,we suggest that the standard normal cumulative distribution function used in VPIN model should be replaced by sigmoid function family commonly used in neural network and genetic algorithm to determine the optimal parameters,so as to reduce the classification error of active trading volume.Sigmoid function can retain most of the advantages of the original basis function,at the same time,it has a more accurate description of the price differences between the standardized trading basket,approaching the true distribution of the standardized price differences as far as possible,so as to achieve more effective improvement.In the aspect of VPIN model and volatility prediction,we rank all data according to the CDF of VPIN and its change rate,and calculate the mean value of Volatility Measurement Index in each subgroup,in order to judge whether the CDF of VPIN and its change rate are the important factors affecting the abnormal volatility of stock index futures.It is found that not only VPIN CDF and its change rate have a stable positive correlation with the volatility of stock index futures,but also the change rate of VPIN CDF enlarges the positive correlation between VPIN CDF and the volatility of stock index futures.