A Research On The Applicability Of The Game Theory CAPM Model In China’s A-share Market

One of the most interesting puzzles in financial economics is the equity premium puzzle,which observes that over the past century equity returns tend to be higher than bond returns.Many studies have shown that the stock premium is around6%.The problem is that the premium is so large that it cannot be explained by investors’risk aversion.This article proposes an alternative explanation for the mystery of stock premiums,which is that attribution of stock premiums to speculation.Based on the effective exponential hypothesis and game theory CAPM model proposed by Glenn Shafer and Vladimir Vovk(2016),this paper defines the specific method of time projection,and compares the performance difference of the model before and after time projection,which explains the rationality of our time projection selection.At the same time,this paper argues that due to the particularity of the Chinese market,the assumptions of the original model are too strong,and there are problems in the applicability of the Chinese market,and it is necessary to make certain adjustments to the model,and then analyze the difference between the adjusted model and the original model in the Chinese market,indicating that the model adjustment has certain rationality,and finally comparing the advantages and disadvantages of the adjusted model and the traditional CAPM model,indicating that the adjusted game theory CAPM model has certain applicability in the Chinese market.This paper uses the CSI 300 Index and CSI 300 constituents from January2005 to July 2022 and 5-minute high-frequency data of yield as the research object.The article first verifies the source of volatility and finds that the variance ratio of the daily yield from Friday to Monday and the daily yield from Tuesday to Friday is about 1.5,which is much less than 3,indicating that the income fluctuation in the stock market is largely caused by the transaction itself,and speculation is usually accompanied by a large number of transactions,which in turn shows that it is of practical significance to study the mystery of stock premium from the perspective of speculation.Further,based on the basic assumption of the article,the paper projects time in a fixed small increment ofΣ_t/n squared by the square of the yield,based on the basic assumption of the article.In this paper,n=200 to 4000,the step value is equal to 100,and the performance ability of the effective exponential hypothesis model before and after time projection is compared,and it is found that the performance ability of the model has indeed been significantly improved after time projection,indicating that this paper has certain rationality for the selection of time projection mode.On this basis,based on the projected yield series to verify the effective index hypothesis EIH,this paper finds that no matter how the size of n is adjusted,the conclusion of the original effective index hypothesis EIH cannot be obtained,indicating that the original EIH model has poor applicability in the Chinese market.Further,this paper refits the drift termμaccording to the projected index yield data to obtain the adjusted EIH model,and finds that compared with the original model,no matter how the value of n changes,the explanatory ability of the adjusted model on the exponential premium scale has been significantly improved,indicating that the assumptions of the original model are reasonable to a certain extent,and at the same time,this paper tests the normality of the residual after fitting and finds that no matter what the value of n is,the fitted residual term generally obeys the normal distribution.It is explained that the refitting of the model does not violate the original basic assumptions,and has theoretical basis and practical significance.This paper finds that with the increase of n,the projection increment gradually decreases,and the model’s explanatory ability for the exponential premium scale shows a trend of gradually getting better and then deteriorating,and there are obvious inflection points,and it is found that when n is selected between 800 and1600,the adjusted EIH model can have better performance ability.According to the EIH model after refitting the trend term u,the adjusted game theory CAPM model can be obtained,according to the best selection range of n determined earlier,in the study of the premium scale of CSI 300 constituent stocks,this paper selects n=800 to 1600,and the step value is 200,which verifies that the adjusted game theory CAPM model has better explanatory ability for the premium scale of CSI 300 stocks than the original model,and compares and studies the performance ability differences of different n models.The value of the best n is800,1000.Finally,based on the best n=800,1000,this paper compares the difference between the adjusted game theory CAPM model and the original traditional CAPM model on the premium scale of Shanghai and Shenzhen 300 stocks from 2005 to2022,and finds that the adjusted game theory CAPM model performs better.In addition,the data of 2008 are eliminated,and the robustness test of the conclusions of this paper is carried out from the new sample period,and it is found that the conclusions of this paper have certain robustness.This paper explains the stock premium from a new perspective,that is,from the perspective of speculation,does not make any assumptions about investors’risk attitude,defines the specific time projection method on the basis of the original effective index hypothesis model,and makes certain adjustments to the original model based on the actual situation of the Chinese market,which provides a new research idea and perspective for the mystery of stock premium.