The Effect Of Market Sentiment On Option Pricing

In the development process of asset pricing theory,Black and Scholes proposed the famous BS option pricing model,which laid the foundation of option pricing theory,and the model is significant for the historical development of option pricing.However,with the development of option pricing,a large number of studies have shown that there are discrepancies between the option prices derived from the BS model and the real option prices in the market.The assumptions behind the BS model are too strict.The theory assumes that investment agents are perfectly rational and the market is complete,and even if there are some irrational investors in the market,the full arbitrage of the market will compensate for the price deviation caused by the irrational investors,and the irrational behavior of investors can be ignored at this time.However,behavioral finance research shows that investors are irrational and their investment decisions are influenced by emotions,and the limited arbitrage of the market cannot eliminate the pricing bias brought by emotions.Therefore,if the impact of investor sentiment on option prices is considered in option pricing,it can help to improve model pricing efficiency and reduce estimation bias.In this paper,two main aspects are carried out.First,a sentiment indicator is constructed with stock review texts,and the two major methods,machine learning model and deep learning model,are used to identify stock review texts and convert them into investor sentiment.Second,we construct an option pricing model based on market sentiment,numerically simulate the effect of market sentiment on option prices,and empirically test the improvement effect of the option pricing model with the addition of market sentiment factors.Specifically.First,the market sentiment index is constructed by collecting investor comments from 2018 to 2019 through web crawling of "Eastmoney",and then converting them into investor sentiment by two methods: machine learning model and deep learning model,and evaluating the recognition effect of the two methods.It is found that the classification effect of the deep learning model is more excellent compared to the machine learning model,in which the long short-term memory model(LSTM)of the deep learning model can improve the recognition accuracy to more than 92%,which is significantly better than the classification accuracy of other literature.Second,the option pricing model based on market sentiment is constructed.Based on the existing BS model,we relax the assumption of investor irrationality and construct the derived option pricing model based on the assumption of investor irrationality by studying the effect of market sentiment on the discount rate.After that,the pricing effect of the improved option model is measured,and it is concluded that the improved BS option pricing model can reduce the theoretical price and actual price errors for both call and put options,and the correlation between the theoretical price and actual price is further improved.This paper constructs a text sentiment indicator,empirically analyzes the effect of sentiment on option prices,derives an option pricing model based on market sentiment,and conducts a model improvement test to demonstrate the systematic influence of market sentiment on option pricing,which helps investment agents understand option price behavior and thus avoid risk.