Research On Stock Price Prediction Based On Improved Particle Swarm Optimization Neural Network

With the deepening of China’s reform and opening up,China’s economy has entered the train of rapid development,and its GDP has jumped to the forefront of the world.Various financial methods and financial products have dazzled investors.In the capital market,various financial products such as stocks,funds,futures and bonds have emerged in an endless stream.Due to the stock limit,the trading day is extended and there is no threshold limit for starting capital.Therefore,in the capital market,the stock of listed companies is favored by the majority of investors.But due to the rise and fall of the stock closed a variety of factors.This includes international capital peripheral stock market,international major news emergencies,national policies,institutional operations,listed companies operating conditions,etc.Due to the influence of many factors,it is even more difficult for individual investors who want to make money in the stock market.At this time,a number of stock prediction methods come into being.Compared with traditional time series method,security analysis method and other traditional stock prediction methods,artificial intelligence method is superior.So stock market investors are more likely to believe in artificial intelligence.Among them,the neural network algorithm to predict the changing trend of stock trading prices is more prominent.Because the parameters of neural network algorithm are less simple,it is easy to implement.So the neural network algorithm has been widely used in various fields.However,the neural network also has its shortcomings,such as improper selection of weight and bias,which leads to premature loneliness of the algorithm,and thus cannot reach the global optimal.The predicted value that approximates the actual result is not obtained.This paper presents an improved particle swarm optimization algorithm to optimize BP neural networks.According to the model,the capture obtained from the simulation experiment on the historical data of the stock can be well fitted to the actual stock trading price.It can help investors reasonably allocate funds,avoid risks and expand returns.The main research work of this paper includes the following two aspects:Particle swarm optimization algorithm is improved to solve the problem that the traditional particle swarm optimization algorithm is easy to fall into the local optimal value.Make targeted improvements to it.The inertia weight is adjusted dynamically.In the early stage,the inertia weight is large,which is suitable for global search.With the operation of the algorithm,the inertia weight decreases in the late stage,which is suitable for local search.In addition,the compression factor is added to the velocity update formula,which can effectively search different regions and obtain the best solution of high quality.According to the improved particle swarm optimization algorithm BP neural network to establish the corresponding model and simulation experiment,choose the listed company Xiangcai stock trading data as the historical data as the experiment sample.According to the experimental results,it is found that BP neural network optimized by the improved particle swarm optimization algorithm has a good credibility in the prediction of stock trading price.