Research On Program Trading Using The Neural Network Prediction Model Based On QPSO Algorithm

This research is devoted to construct a time series prediction model by means of artificial neural network and intelligent optimization algorithms to analyze and forecast the financial time series effectively.Further,took the accurate prediction result of the model as core to program a trading strategy and complete simulative programming trading experiment.The main contribution and conclusions of this research are as follows:First of all,the thesis described the origin,development,improvement and application of particles swarm optimization in detail,especially simulated annealing-particle swarm optimization and quantum-believed particle swarm optimization.Then,this research constructed the time series prediction model using the artificial neural network combined with SAPSO and QPSO respectively,and made a comparison between them from qualitative and quantitative aspects.Analysis result showed that both the conciseness,simplicity of parameter setting,easy understanding of algorithm itself or expansion ability,prediction ability and operation efficiency,the neural network prediction model based on QPSO algorithm is superior to the model based on SAPSO algorithm.Further,this research used the neural network prediction model based on QPSO,which has better prediction ability,to carry on modeling,simulating and predicting the time series of HaiTong security.On the basis of prediction results,through the form of simulated trading,firm offer operations on PMS platform of WIND and sustained profit to prove the validity of the model and its applicability in programming trading.Finally,ideas and outlook of future research were proposed after reviewing the the whole process of the experiment research and referring to the problems it once occurred.