Prediction and Analysis of Cotton Futures Prices based on LSTM Model

Abstract

In the context of global economic integration, the cotton futures market, as an important component of the agricultural futures market, not only affects the economic benefits of many enterprises in the cotton industry chain, but is also closely related to the macroeconomic situation due to price fluctuations. In view of the nonlinear and unstable characteristics of cotton futures prices, in order to better predict cotton futures prices accurately, this paper selects CCINDEX3128B as the basic model, collects relevant price data from Zhengzhou Stock Exchange from January 2, 2019 to December 18, 2024 as the in-depth research object, and innovatively uses Long Short Term Memory Network (LSTM) model for systematic analysis and prediction. In the research process, comprehensive data related to cotton futures prices was first collected, covering various influencing factors mentioned above, and strict data cleaning and preprocessing were carried out to ensure the accuracy and usability of the data. Subsequently, the processed data is inputted into the constructed LSTM model for prediction. In the process of model training and prediction, key indicators such as mean absolute error (MAE) and coefficient of determination were comprehensively analyzed to evaluate the predictive performance of the model. The results showed that the LSTM model exhibited good goodness of fit in cotton futures price prediction, with a MAE index of 117.45718 and an RMSE index of 153.78997, both of which were at a low level. This indicates that the error between the model's prediction results and actual values is small, and it can accurately capture the fluctuation trend of cotton futures prices, providing a high reference value for cotton futures market participants to make more scientific and reasonable decisions in market trading.