| People continue to explore how to apply mathematical theory to control financial risk with the development of the financial industry. As a result, quantitative investment was born. It refers to the use of quantitative methods on the market data for statistical analysis, and then lead to investment decision. Operators can use a variety of scientific method in quantitative analysis for human or computer while making statistical analysis depending on market data in the investment strategy. The development and application of quantitative investment overseas is relatively mature. Quantitative investment has been more and more accepted by investors because of stable investment performance and expanding market size. The development of quantitative investment will accelerate the advance of market while the domestic is at the stage of development. In this context, we write the computer code to realize the program trading, and develop a program trading system, relying on the hot leading edge of academic theory and on the on the basis of practice of trading.In this paper, we make trading strategy and realize the automatic transaction using non-parametric statistical methods and the linear time-varying semi-parametric model. First of all, a nonparametric regression model is established. We put all the technical indexes of market into the non-parametric regression model variables, and put the futures price into the dependent variable. We use the LCLS (Local-Constant Least-Squares) estimator in the non-parametric regression to select variables which have the statistical significant effect on the dependent variable. Then we use the LLLS (Local-linear Least-Squares) estimator in the non-parametric regression to identify variables which have linear or nonlinear effect on price of the futures. Consequently, we get the semi-parametric model. Next, we make pricing forecasting based on semi parametric model which is obtained. Then we float up and down in a certain range to obtain the forecast interval. When the price is above of the interval, it should be sold because of an excessive price. When the price is below of the interval, it should be bought because of a lower price. When the price is in the interval, we should not trade because it may have no arbitrage opportunity. Finally, connect the R procedure and the TB procedure to realize the programmed automatic transaction. Before actual transaction process, trading strategy must be measured back to the historical data and be modified the strategy according to the results of test performance which reflects a statistical thinking. Excellent trading rule is built on the basis of statistical analysis of data, and excellent trading behavior is built on the basis of accurate prediction. According the hot leading edge of non-parametric method and the higher fitting degree of linear time-varying semi-parametric model, we successfully developed a system of quantitative trading strategies. This is a practical application of statistical theory and mathematical models. For scholars, it can strengthen confidence in theoretical innovation that the leading edge of academic. For investors, it can catch the arbitrage opportunity which can also enhance market activity. For the market, it can strengthen the liquidity of the capital market and accelerate the development of market maturity. In addition, for supervision department, the model can be used to monitor the market after putting the macro factors into the Semi-parametric model.Based on the perspective of large data, we use the non-parameter statistical method for data mining, finding out the data set which required. Therefore, the innovation of this paper lies in the new technology. What’s more, its application is very extensive. Non-parametric statistical method and semi-parametric model not only can be used to the market as the basis of trading strategy, but also can be used to monitor the futures market operation in order to estimate the impact effect of macroeconomic factors on the futures market. |
PDF Full Text Request