Optimal Algorithmic Trading Strategy And Empirical Analysis Based On Market-Maker Model

With the continuous improvement of the financial market and the rapid development of electronic trading systems and communication technology,the traditional manual trading method has gradually changed to electronic automated trading.Programming trading has become the majority of relevant workers with its relatively fast calculation efficiency and more accurate results.With the increasing attention of investors to risk and the continuous development of risk diversification,the role of programming trading has become increasingly important.It has gradually developed into an irreplaceable trading model in capital market.Algorithmic trading is a very important part of programming trading.It guides the trading timing,price and quantity of orders through pre-set trading strategies without manual intervention,and automatically adjusts the instructions according to changes in the market.Algorithmic trading is currently developed as one of the methods that investors often use.The main features are to reduce transaction costs,control market impact costs quickly and effectively,hide trading intentions and improve order execution efficiency.In the securities market,when institutional investors conduct securities trading of large orders in one single transaction,it will have a huge impact on the market,which may cause the price of securities to change in an unexpected direction due to the limited liquidity of securities.However,if the order is divided too tiny,the transaction time will increase,and the price is more likely to change.Based on this problem,this thesis proposes an optimal algorithmic trading strategy based on the market-maker model.In the implementation of the algorithm,this thesis has made some innovations on the basis of traditional algorithmic trading:we combine high-frequency trading with algorithmic trading and propose to consider only sell limit orders and market orders in the framework of high-frequency trading market-making strategies.Under the goal of maximizing wealth,we set the target value function,use the dynamic programming principle to introduce the quasi-variational inequality,and solve it by the finite difference method.According to the result of the strategy,the ordering strategy A =(?make,?take)of different inventory,different depths and different price differences at each time point in the transaction is guided,where the random control ?make guides the limit order and the pulse control ?take guides the market order--to complete the entire optimal algorithmic trading strategy.According to the optimal algorithmic trading strategy constructed in this thesis,we use the high frequency data of active species in three different markets in 2018 to conduct an empirical analysis of the algorithm.Through the comparison of the wealth value after the completion of the large-order transaction at the end of the trading day,we prove that the optimal algorithmic trading strategy based on the market-maker model proposed in this thesis can improve the efficiency of order execution and reduce the impact cost of the market,which is more effective than the common algorithmic trading strategies such as VWAP and TWAP.