Machine Learning-based Dynamic Instrument Price Model

It has been a focal problem in the field of economics and computer to establish a dynamic bargaining system, which can simulate the universal phenomena in economics and open out the general bargaining rule of human beings, using agent techniques and machine learning theory for a long time.An amendatory dynamic bargaining system based on machine learning is presented in this dissertation. First, we establish a dynamic bargaining system using genetic algorithms (GADBS) according to Oliver's theory, and adopt decimal code instead of binary code in order to prevent variables from going beyond limitative scope. Furthermore, we discuss the drawback of GADBS and mend it in the dynamic bargaining system based on machine learning (MLDBS). In MLDBS, we analyze the situation in detail that how does agent extract characteristics from its opponents, and a BP neural network is used as classifier to identify the opponents. Finally, we compare the experimental datum of GADBS with the experimental datum of MLDBS.The result of our experiment shows that the agent in MLDBS not can only identify its opponents successfully but can change its strategy in term of different opponents in bargaining process.