Research on the application of time series ARIMA model in trade strategy

Abstract

Market trading has always been a popular choice for bold investors, but market trading doesn't rely on pure luck. More need for daily experience and data. In this article, we will model the random forest algorithm for a specific situation and find the best strategy. Then it is proved that the model can provide the best trading strategy, which can be understood as proving the fit of the model established above. Using the real value of bit coin and gold prices in 5 years, establish a time series ARIMA model, analyze the smoothness and pure randomness of the series.Then get the ARIMA model fitting of bit coin and gold, indicating that the prediction model is very suitable. For the dynamic programming model, this paper adopts a genetic algorithm for the judgment of the optimal buy-sell node, which accelerates the convergence speed. To understand how the trading price affects the strategy and the results, sensitivity analysis is also done in this paper. Considering giving a certain perturbation to the results and giving a perturbation margin of 5% to the buying and selling fees respectively, the algorithm is iterated to calculate the sensitivity of the investment strategy to the trading cost and analyze how the change in the trading cost would affect the buying and selling strategy.