Study On The Impact Of Investor Sentiment On Soybean Futures Prices

Agricultural commodity futures reflect market expectations through price information,providing a scientific reference signal.In China,soybean futures trading volume ranks high among agricultural commodity futures markets.However,significant fluctuations in soybean futures prices prompt market participants to increase their demand for hedging.Research on soybean futures prices contributes to risk mitigation and revenue enhancement for China’s soybean production and operations.With the development of mobile internet,an increasing number of individual investors search for investment-related information online.Although individual investors invest in their personal capacity,they generally engage in discussions on social platforms as a reference for investment.Among various platforms,this study selects the posts from East Money’s forum to construct an investor sentiment index.This paper uses soybean-related posts from East Money’s stock forum as the research sample,employing text analysis techniques to mine investor sentiment in the post data.It constructs a weekly bullish indicator,investor attention index,and investor opinion convergence index to prepare for subsequent research and modeling of the correlation with soybean futures prices.Multivariate regression and Granger causality tests are used to explore the relationship between investor sentiment and soybean futures prices.The three sentiment indicators are used as features,and the rise and fall of soybean futures closing prices after post publication are used as the target variable.The accuracy of soybean futures price prediction is tested using random forest and support vector machine machine learning models.The importance of the three sentiment indicators on soybean futures prices is assessed using a random forest model.The study reveals that:(1)There is a significant relationship between investor sentiment and soybean futures prices.Both the bullish indicator and investor attention index show a positive correlation with soybean futures prices,meaning that when these indicators rise,soybean futures prices may also increase.Conversely,the investor opinion convergence index is negatively correlated with soybean futures prices,indicating that when this index rises,soybean futures prices may decrease.(2)The influence of bullish indicators,investor attention index,and investor opinion convergence index on soybean futures prices is unidirectional.This suggests that these indicators may affect soybean futures prices,but soybean futures prices will not,in turn,affect these indicators.(3)Using feature importance ranking of random forest regression and classification models,the study analyzes the importance of various investor sentiment indicators on soybean futures prices.It finds that among the sentiment indicators,the investor attention index has the closest relationship with soybean futures price fluctuations,followed by the bullish indicator,while the importance of the opinion convergence index is the lowest.This paper utilizes text analysis techniques to classify sentiment tendencies and construct investor sentiment indices,reflecting investor subjective emotions that traditional methods cannot capture.It adopts machine learning algorithms more suitable for sentiment classification tasks in the financial domain to improve classification accuracy.Compared to methods that use sentiment dictionaries to build sentiment classification,this approach is better at capturing emotional changes in financial texts.