Research On Market Maker Quotation Strategy Model With Inventory Penalty

In the financial market,market makers,as important participants,can enhance market liquidity,maintain price stability,and improve the timeliness of transactions.Reasonable quotation strategies and models are needed to guide decision-making.The goal of market makers' strategies is to determine optimal bid and ask quotes,maximize the transaction volume with the largest possible spread,and ensure that the inventory is as small as possible under the conditions of limited inventory and risk tolerance.This thesis has conducted a comprehensive and systematic research on the market maers' quotation strategies and models,covering both stock and option market,and provides a complete and detailed description of market maker issues,from theoretical and practical perspective.In chapter 2,we introduce the theoretical knowledge related to the market making(MM)problem.Firstly,we depict the problem of MM and limit order books.Secondly,we review several traditional MM strategies.Finally,the stochastic optimal control theory related to the solutions of the MM model is introduced,as well as how it is applied to solve this problem.In chapter 3,we propose stock MM strategies and models.In dealing with the inventory risk of a single stock market,we both set the upper and lower bounds of inventory and increase the final inventory penalty parameter ?.The new MM model is solved under the CARA utility function.We can prove the existence of the solution and obtain an analytical solution for bid and ask quotes.Further,in the real trading situation,to solve the MM problem of stocks with up or down trends,MM strategies based on technical indicators SMA and RSI are proposed.Market makers at this time are more inclined to unbalanced quotation,and more limit orders in their preferred direction.For the MM problem of multiple stocks,hedging the risks between stocks through factor combination,considering both the profitability and the stability in the long run,a multi-factor model based MM strategy is proposed.Finally,using the real data of China's A-share market for backtesting analysis,we can conclude that the strategy proposed in this chapter can achieve a higher return and a wider range of application.In chapter 4,we propose option MM models and strategies for single and multiple options.To address the inventory risk,we construct a delta-hedged portfolio and consider the CARA utility function with inventory penalty.The framework for modeling a single option MM problem integrates option pricing theories and option risk Greeks,combined with stochastic optimal control theory,and we can prove the existence of the solution and obtain an analytic formulas for bid-ask quotes.For MM of multiple options,higher-order Greeks need to be considered.We propose a.new framework to provide quantities for each quote,with constrains on the portfolio Greeks.Then the MM problem is transformed into a.linear programming problem,which can be solved to obtain multiple option MM strategies.Based on the trading data of 50-ETF and its European options,we develop strategies based on the analytic quoting formulas,and perform empirical studies to demonstrate the performance and advantages of our method.In chapter 5,we construct MM strategies based on deep reinforcement learning(DRL),using efficient SAC algorithm.Firstly,the relevant theories of applying DRL to the MM problem are introduced.Secondly,a general reinforcement learning framework and a deep learning network for MM problem are given,and an intelligent market maker's training process based on SAC is designed.Our control of the inventory risk is not only in the reply of the inventory status,but also in the design of inventory penalty on the reward function.In the simulation analysis,comparing with Avellaneda-Stoikov model,then we can see that the SAC inventory quotation strategy performs more flexibly on quotation and inventory control.Finally,without relying on any market and price assumptions,we do backtesting analysis on the historical minute data of Hundsun Technologies Inc.The results show that market makers based on SAC algorithm can well complete the task of simultaneous decisions on giving bid-ask quotes and the quantity of quotes,while achieving positive revenue and control inventory risk.