Optimization Of Turtle Trading Model Based On LSTM Neural Network

With the development of China's economy and quantitative trading technology,more and more investors are participating in quantitative trading.The turtle trading model is one of the well-known quantitative trading models.However,in recent years,the profitability of this model in China's futures market has shown a downward trend year by year,and investors lack effective theoretical tools to improve the model.To solve the above problems,this paper proposes an optimized turtle trading model based on LSTM neural network.This model improves the channel breakthrough system of the traditional turtle model.At the same time,based on the DTW dynamic time warping algorithm,an algorithm index is proposed to detect whether the model is invalid.The main contributions of this article are as follows:1)Add the technical factors of the stock market to the turtle model,and perform factor screening to find the appropriate factors.There are more studies on the domestic stock market,but less on the futures market.In the field of stock quantitative investment research,multi-factor model research is very mature and widely used in quantitative trading.The turtle trading model does not introduce multiple factors.This article proposes to add the technical factors of the stock market to the turtle quantitative model to improve the profitability of the model.2)Proposed a turtle trading model based on LSTM neural network optimization.This article uses the Long and Short-Term Memory Network to build a trend prediction system,which replaces the channel breakthrough system of the original traditional turtle trading model.Experiments show that under the same experimental environment,the optimized model's profitability in rebar,natural rubber and zinc futures varieties all exceed the traditional model.3)Based on the DTW dynamic time warping algorithm,an algorithm index is proposed to detect whether the model is invalid.The quantitative model is not long-term effective.In order to detect whether the optimized quantitative model fails,this paper creatively combines the DTW value with the model's return rate as a judgment index for detecting whether the model fails.Experiments show that this indicator improves the overall profitability of the model by more than 20%.Finally,the method proposed in this paper expands the application field of the LSTM algorithm and solves the problem of the gradual failure of the traditional sea turtle model in recent years.In addition,in order to detect whether the model is invalid and ensure the profitability of the model,this article uses the dynamic regularization algorithm combined with the model's rate of return as a measurement index.