Empirical Research On Forecasting Stock Returns Based On Realized Skewness

Capital asset pricing is the core of the study of financial assets research.The theoretical and empirical research on capital asset pricing has never stopped.Among them,the most pioneering research is Markowitz’s mean-variance model,which innovatively put forward the concept of risk and return,and converts the concept into mathematical statistics.Subsequently,Sharpe(1964),Lintner(1965)and Mossin(1966)proposed CAPM to further study the return on assets and risks.However,the mean-variance model assumes that returns follow the standard normal distribution,and CAPM merely considers market volatility risks,which are different from the distribution of actual stock returns.The distribution of stock returns actually has the characteristics of sharp peaks and thick tails.Therefore,the skewness and kurtosis should be considered.At present,research on the impact of higher moments on capital asset pricing is increasingly abundant.Among them,Kraus and Litzenberger(1976)believe that co-skewness is the decisive factor affecting the cross-section of stock returns;Fang and Lai(1997)first proposed a fourth-order moment asset pricing model,showing that the stock’s excess return is also related to market kurtosis and market skewness.Merton(1980)pointed out that as the frequency of data sampling increases,volatility can be measured accurately.Many subsequent studies of actual volatility have adopted this approach,using the sum of squared daily returns to construct the actual volatility.For the study of the relationship between higher moments and stock returns,Amaya extended this method by calculating the actual skewness using the cubic power of intraday returns,extending the existing concept of actual volatility.This method of calculating actual skewness uses a model-free analysis method.Compared to low-frequency data,the high-frequency transaction data contains a lot of rich information,no complicated modeling process and parameter estimation are needed,and it does not exist an estimation error is formed in the parameter estimation process.Therefore,this article starts from the realized skewness of the Shanghai Stock Index,uses Amaya,Christoffersen,Jacobs,Vasquez(2015)to construct the realized skewness method,and uses the 5-minute high-frequency trading data within the day to calculate the intra-day rate of return,intra-day volatility,and intra-day Realized skewness.Then using the realized skewness as the independent variable,and the return rate of the stock market in the next month as the dependent variable to investigate whether the realized skewness significantly affects the future return rate of the stock;Second,after exploring the relationship between the realized skewness and future returns,the article will further conduct out-of-sample predictions to investigate whether the realized skewness still has a significant forecasting ability,and will also conduct a test of economic significance.That is to explore whether the realized skewness can effectively improve the return of the investment portfolio.Finally,the article conducts a robustness test on the realized skewness and the future return of the stock market.Through a series of empirical studies,the article has obtained the following main results:(1)The results of the study show that there is a significant negative correlation between the realized skewness of China’s stock market and the future return of China’s stock market.That means in the stock market,when the realized skewness has been reduced,the future return of the stock market will be even higher.The estimated parameter value of the realized skewness is-0.0524;(2)The realized skewness has passed the in-sample and out-of-sample prediction tests,which further confirms the predictive ability of the realized skewness;(3)The realized skewness,as the main factor influencing asset pricing,is tested at the level of economic significance,which produces significant economic results.The annual management fee that investors are willing to pay in order to obtain an investment strategy based on the realized skewness is 323 basis points.At the same time,the Sharpe ratios of portfolios based on realized skewness is 0.1929,which is also relatively high.Therefore,in terms of economic significance,the realized skewness has a certain influence on the returns of investors’ future portfolios;(4)Contrast the unconstrained univariate regression forecasting model,the forecasting research model based on the explanatory factor constraint method has become more significant.The R2 improved by 0.1%.