| The development of behavioral finance has provided a theoretical basis for the impact of investor sentiment on asset pricing.Finding appropriate indicators to reflect investor sentiment and studying the impact of investor sentiment on financial assets is of great significance for enriching and improving the research on the impact of investor sentiment on financial assets.At the same time,more and more investors are used to expressing their views on the investment object in public forums,which may not only be affected by the price fluctuation of the investment object,but also may have an impact on the price of the investment object,Therefore,it is necessary to study the relationship between investor sentiment and underlying asset prices.In the current research on the relationship between investor sentiment and financial assets,it is common to conduct principal component analysis on multiple indicators containing investor sentiment,and use machine learning methods to analyze text semantics,and then study the relationship between financial assets such as stocks and investor sentiment.However,in the field of futures investor sentiment research,There are few studies on the impact of futures investor sentiment on futures prices using text analysis methods.This paper divides the comments of glass futures investors into three categories,trying to analyze the relationship between investor sentiment and glass futures price and volatility under different market conditions,and explore the asymmetric impact of investor sentiment on glass futures price and volatility.First of all,this paper crawls the investor comments of the Glass Futures Bar of Oriental Wealth Futures,then classifies the investor comments by date,and then classifies the investor’s comments on the day by name to avoid multiple repeated comments by the same investor on the day.After sorting out the comments,this paper will integrate the comments to preprocess the text,conduct word segmentation and stop word removal for the Chinese text,and then use the CBOW model in Word2 Vec to train the corpus.After getting the training set,this paper first manually sorts out the positive semantic words and negative semantic seed words of investor comments,and then puts them into the result set trained by CBOW model to get the synonym set of different semantic words.After obtaining different emotional semantic word sets,this paper conducts semantic analysis on each investor’s comment,classifies investors into bullish positive emotional investors,bearish negative emotional investors and wait-and-see neutral emotional investors,and makes statistics on the number of investors of three types every day.Secondly,the SVAR model of investor sentiment and closing price difference of glass futures is constructed.This paper first analyzes the relationship between the number of investors with different emotions and the closing price difference of glass futures under different market conditions through Granger causality test.Then we analyze the impact mechanism between the number of investors with different emotions and the closing price difference of glass futures under different market conditions through impulse response and variance decomposition.The results show that:(1)In the horizontal trading period of glass futures,the number of investors with different emotions has little influence on the closing price difference of glass futures,and the closing price difference of glass futures has little influence on the number of investors with different emotions.In the horizontal trading period,because the price of glass futures fluctuates steadily up and down within a certain price range,it will not cause a greater response from investors.Therefore,the change in the closing difference of glass futures cannot be transmitted to the number of investors with different emotions,and the change in the number of investors with different emotions will not cause a significant change in the closing difference of glass futures.(2)In the rising period of glass futures,the number of investors with different emotions has little influence on the closing price difference of glass futures.The comments of glass futures investors on glass futures lack the transmission mechanism to the closing price difference of glass futures in the rising period.The closing price difference of glass futures is mainly affected by other factors.The closing price difference of glass futures has a continuous positive impact on the number of investors with different emotions,and the behavior of glass futures investors has the hot spot effect.(3)In the falling period of glass futures,the number of negative investors has a certain positive impact on the closing price difference of glass futures,while the closing price difference of glass futures will have a small negative impact on the number of negative investors.Thirdly,this paper explores the interaction between glass production and application factors and glass futures prices.This paper finds that there is no significant relationship between the daily fluctuation of the closing price difference of glass futures and the closing price difference of soda ash,steam coal,crude oil futures,NYMEX natural gas futures,and the real estate index of China Securities Regulatory Commission.Finally,Taking the closing price of glass futures as the dependent variable and the number of investors with different emotions as the independent variable,the GARCH model is constructed.The results show that the more active investors,the smaller the volatility of glass futures prices.The more negative investors and neutral investors,the greater the volatility of glass futures prices. |