Price Forecasting Of CSI300Futures Based On BP Neural Network

In the1970s,the changes of international economic situation brought the reform of the financial system.So many Financial futures derivatives like as Interest rate and currency futures emerged In this complex context.At the same time,the stock index futures boarded the stage of history and became the backbone of the financial derivatives in futures market after a period of rapid development.The road of the development financial market is not flat in our country.In the early1990s, our country carried out the pilot of the financial futures trading,but brought very serious consequences because of the lack of supervision and other reasons. After several considerable reorganization,developed slowly.As China’s CSI300futures officially listed in April16,2012,China’s financial futures market has entered a phase of vigorous development.The introduction of stock index futures proclaimed the passing of an old era, also declared a new era’s coming.This is a signal that the financial markets tend to be perfect and a big step we have taken In the financial reform.Because the market started late and the basis was so weak,so our financial futures markets can not be compared with developed countries.At present,only CSI300futures and five-year bond futures are traded on the market.As the only type of financial futures, Chinese stock index futures has been listed for four years. Since2014, although trading volume makes up less than10%of the entire futures market, the turnover has been ahead of other futures.Since listing, the cumulative volume of futures contracts has reached over455million, increased from the initial of2702to now nearly140,000, and trading volume once hovered around800,000.Prosperity of the financial market is inseparable from the word risk. If the risk is not controllable, the whole financial market and the real economy will suffer a fatal blow.We all know that stock index futures has an irreplaceable role in the fight against risk, but in all aspects of that influence is also indisputable. The index has slowly infiltrated into our financial system to constantly affect our investment and financing environment. For the entire country, it improves stability of the entire financial system and for us individual investors,it provides good hedging and investment instruments.There are four kinds of index futures contracts, regardless of the volume or the amount of the transaction, current month contract occupies a very important position, so data in this article are selected from the current month contract, known as continuous data of current month, a total of708data from August8,2011to July11,2014.We take data from the last28days as a validation sample, and the previous data as training samples, and then predict prices via BP neural network. Specific methods consist of three experiments. First Experiment is price predict based on a single variable, the closing price. We forecast the closing price of the eleventh day based on closing prices of the first ten days.Second experiment adopts other variables, such as futures trading volume, the subject that CSI300index’s closing price, the Shanghai Composite Index and Shenzhen index, and uses data of the first five trading days and variables of the fifth day to predict the closing price of the sixth day. This design allows input of networks information become more abundant, making the prediction accuracy and convergence time of the network improve.Third Experiment introduces wavelet transform for data noise reduction processing, and then predict via the network, making the network faster compared with first experiment and second experiment, and essentially optimizing the stability of the network. Through such gradual research, we can improve prediction accuracy, step by step, and the stability of the network can be gradually improved.The innovation of this paper is as follows:The innovation of the study samples. In this paper, the research range is from August8,2011to July11,2014.From the time dimension, the data is very new and the research of the price is the most current.2、The innovation of the neural network.In the first experiment,the adjustment for the network input data make the network input nerve node into10.In the second experiment,there is a new matrix input which introduces other relevant variables and uses the five-day closing price plus all other variable data the day before to predict the closing price that makes the prediction accuracy increased from13.516to11.856.This is a great progress.3、The innovation of problems studying.The goal of this paper is not only based on the forecast of each trading day’s closing price, but also the trend of the stock index.In the structure of this paper,the first chapter introduces the research background, the significance of the selected topic and the research status at home and abroad.The second chapter expound some related theory about BP neural network and wavelet analysis to provide a theoretical basis for the empirical research.The third chapter is the empirical research of futures price forecast.The fourth chapter is the conclusion of this paper, as well as the innovations and shortcomings of this paper.The conclusion is:the forecast of the closing price of the stock index futures can be realized through massive experiments. With constantly deepening the experiments, we can find out that the prediction accuracy is continuously improved. It achieves high precision of fitting and forecasting effect. And through three experiments and the data, we can say:within a month trading-days after June3,2014, there is a rising trend of the stock index futures prices. This conclusion is proved.