Research On Trading Strategy Based On Time-Varying Quantiles Model Of Non-Parametric

In the last century, with the rapid development of computer technology and financial mathematics, program trading is gradually replacing the traditional subjective transaction, a lot of strategies are proposed early which based on fundamental analysis and technical analysis, but at quantitative time, More and more investors appeal to trading strategy designed by quantitative model. Due to the imperfections of Chinese financial market system and the lack of short-mechanism, the proportion of program trading in China is far lower than in the foreign mature markets, but after promoting Shanghai and Shenzhen 300 stock index futures, Shanghai 50 stock index futures and CSI 500 stock index futures, The short-mechanism of Chinese capital markets become more perfect, for program trading investors are concerned, it will be a good opportunity to flex their muscles.While program trading is being carried out in full swing in China, the author designs a program trading strategy based on a quantitative model. The author provides investors with a new set of program trading system and wishes investors to extend the idea of designing trading strategy, and apply the non-parametric method to the field of designing program trading strategies.This paper describes program trading systems classification and the entire process of designing program trading systems, next trading strategy designing process in article is in accordance with a flow chart of designing program trading system tightly in order to obtain a rigorous and convincing trading strategies. Then the paper introduces trading strategies designed based on time-varying quantiles model of non-parametric kernel density estimation and examines the ability of estimation and prediction of the time-varying quantiles model. Combining time-varying quantiles model with Bol linger band channel strategy, the author builds a trading range, then designs a counter-trend volatility trading strategy based on the trading range and a set of evaluation system. After designing the trading strategy and evaluation system, the article analyses the trading strategy profitability on the three futures, at the same time taking into account the effects of slippage and stop methods on trading results, and comparing with the three stock market indices corresponding earnings. To compare the performance of the proposed final design of the trading strategy, the article does the same test on the trading strategy based on a neural network model designed for a long time in academia.Finally, the studies have found that slippage is important to program trading, especially high-frequency trading, trailing stop way is better than fixed stop approach. The trading strategy designed by the paper deals with shock market better than the turbulent market. This strategy can resist the risk of the market fell, but associated with certain risks. The cumulative gain of the strategy is far more than the market gains, and income obtained by the trading strategy designed by neural network model, and the risk is relatively small.