Revisiting Fama–French factors' predictability with Bayesian modelling and copula‐based portfolio optimization

Abstract

This study is investigating the predictability of the five Fama–French factors and explores their optimal portfolio allocation for factor investing during 2000–2017. Firstly, we forecast each factor with a pool of linear and nonlinear models. Next, the individual forecasts are combined through dynamic model averaging, and their performance is benchmarked by the best performing individual predictor and other forecast combination techniques. Finally, we use the generalized autoregressive score model and the skewed t copula method to estimate the correlation of assets. The generalized autoregressive score performance is also compared with other traditional approaches such as dynamic conditional correlation model and asymmetric dynamic conditional correlation. The performance of the constructed portfolios is assessed through traditional metrics and ratios accounting for the conditional value‐at‐risk and the conditional diversification benefits approach. Our results show that combining Bayesian forecast combinations with copulas is leading to significant improvements in the portfolio optimization process, and forecasting covariance accounting for asymmetric dependence between the factors adds diversification benefits to the obtained portfolios.

Forecasting and trading Bitcoin with machine learning techniques and a hybrid volatility/sentiment leverage

“…Although SVR sets are inferior to XGB in terms of out‐of‐sample performance, their predictive power is better than other models. The result is in line with several studies that suggest SVR can be a robust prediction tool supporting individual predictors (Sermpinis et al, 2014; Zhao et al, 2019). Finally, although LSTM falls short of the GBDT family, it has more accurate results than the benchmark, in line with recent experiments in BTC prediction (Ji et al, 2019; McNally et al, 2018).…”

“…For both sets of factors (RFE-RF and PCA), we shed light on the predictive power of the GBDT family (XGB and LBM) because of all the negative statistics of the MDM test. Zhao et al (2019) suggested that when the superiority of forecasting models suffers from data-snooping bias, the predictive performance may be attributed to luck. In order to further validate the superiority of the XGB model, we then apply two statistical tools, namely, the SPA test and the MCS test.…”

“…SVR has been widely used in time‐series prediction (Tay & Cao, 2001; Zhao et al, 2019). SVR follows the same principle as support vector machine (SVM) but aims to solve regression problems.…”

Multilayer Network Risk Factor Pricing Model

“…In order to solve the asset pricing puzzle, Fama and French [1] proposed the three-factor model, where the crosssectional variation in average portfolio returns is explained by the excess return of a portfolio, the stock size, and the book-to-market ratio. After this pioneering study, scholars proposed different types of factor models from different perspectives and tested them [3,[13][14][15][16][17][18][19][20]. For example, Lam and Tam [14] investigated the role of liquidity in pricing stock returns and found that the liquidity four-factor model is the best model to explain stock returns in the Hong Kong stock market.…”