Research On Trading Strategy Of Dynamic VWAP Algorithms

With the rapid development of securities market and computer network communication technology,computer-based algorithmic trading has been more and more widely used in securities market.As the most basic algorithmic trading strategy,VWAP(Volume Weighted Average Price)has attracted the attention of institutional investors.VWAP divides large orders into several small orders matching the market volume distribution by forecasting the intra-day volume distribution,so as to reduce market shocks and transaction costs.Therefore,the key of VWAP strategy lies in the accurate prediction of intraday trading volume distribution.Historic VWAP strategy is a static trading strategy,which determines the volume ratio of each interval before the beginning of trading day,so it can not respond to real-time information of the market.The dynamic VWAP trading strategy proposed in this paper incorporates real-time information and can react to real-time market conditions and adapt to market changes more.This paper assumes that the transaction price is an interval average price,so the research on VWAP strategy mainly focuses on the prediction of intra-day trading volume.Dynamic VWAP strategy is to multiply a revised parameter which contains real-time market information on the basis of transaction volume predicted by historical VWAP strategy.Firstly,the function form f of the revised parameters is obtained by statistics of historical data,and then the real-time market transaction data is substituted into the function f to get the revised parameters.At the same time,in order to ensure that each interval can be traded,and the volume of transactions to be completed,not too much or too little,this paper also set a threshold,the whole transaction is divided into dynamic part and historical part.In the empirical test,this paper takes 60 stocks in A-share market as samples.The first 30 stocks are taken as samples.The parameters are determined by numerical experiments,and then the algorithmic trading model is constructed.The last 30 stocks are taken as out-of-sample stocks to test the effectiveness of trading strategies.The average absolute error between the predicted trading volume proportion of VWAP strategy and the actual trading volume proportion of the market is taken as the criterion to test the strategy,called tracking error.124 trading days are selected to study the strategy.The results show that the tracking error of dynamic VWAP strategy is smaller than that of historical VWAP strategy.Dynamic VWAP strategy overcomes historical VWAP strategy,The average probability is 64.41%.At the same time,it verifies that the execution cost of dynamic VWAP transaction strategy is lower,so it can be concluded that dynamic VWAP strategy is more accurate in predicting transaction volume and can effectively reduce transaction cost.The dynamic VWAP strategy proposed in this paper is more effective than the historical VWAP strategy.