Trading Strategy Of Futures Based On Hidden Markov Model

Hidden Markov Model is a mathematical model that identifies the hidden state of the observation vector.By calculating the probability of a hidden state,we can find samples in the historical data that are similar to the current data pattern to predict the future.In this paper,the Gaussian Hidden Markov Model is used and the technical analysis index is included as the feature vector group to predict the future yield of the futures.Then the trading strategy based on the weighted forecast is proposed and the sample back test is performed.Through the parameter sensitivity analysis of the trading strategy model under a single futures variety,it is found that the model has better performance in high frequency trading.Modeling with 15-minute frequency data and conducting off-sample transaction testing yielded higher yields and Sharpe ratios.In the process of sensitivity analysis,a series of characteristics of the transaction model were found.For example,the more transactions of futures varieties with large price fluctuations,the higher the profit.Since only the transaction fee and the minimum lot size are considered in the modeling,and the factors such as the impact cost and the market structure change are not considered,the actual transaction is not necessarily effective.This paper verifies the effectiveness of hidden Markov model in market state recognition and prediction of futures,and proposes a weighted prediction model of futures yield,and discusses the feasibility of trading strategy based on hidden Markov model in futures market and makes reference for the actual transaction.