The Fama-Fench 3-factor model (FF3) is one of the most commonly used models for valuing companies. It is curious to see whether this model can reasonably capture the changes in the treacherous stock market and accurately predict the expected returns of a company in practice. In this article, machine learning is adopted to analyze data from the past five years in terms of the 10-year t-bill return as the risk free rate and the S&P 500 rate of return as the market rate of return to compare and analyses the rate of return of Nestle, Danone, Unilever, Coca-Cola and PepsiCo over the past five years. In the comparative analysis, this essay randomly selected the rate of return for 80% of the days in the past five years as the training set and used the rate of return for the remaining 20% of the days as the test set. Besides, this study uses regression analysis to obtain a theoretical model of FF3, which is then brought to the test set data for testing. In this article, the null hypothesis is that the error is equal to 0 and the alternative hypothesis is that the error is not equal to 0. According to the analysis, hypothesis validation is adopted to check the accuracy of FF3 by analyzing the results of the hypothesis test and using the FF3 derived from the training set to predict whether the error between the resulting data and the actual data is 0.
Read full abstract