TY - JOUR AU - Atance, D. AU - Serna, G. KW - Bitcoin return predictions KW - Conditional skewness KW - GARCHS models KW - Sample skewness T1 - Time-varying expected returns, conditional skewness and Bitcoin return predictability LA - eng PY - 2024/08/01/ T2 - Quarterly Review of Economics and Finance SN - 1062-9769 VL - 96 PB - Elsevier B.V. AB - We employ a GARCH-type model to jointly estimate returns, conditional variance and skewness and show that conditional skewness outperforms sample skewness and conditional and sample variance in predicting future Bitcoin returns. Interestingly, the results show that the relationship between conditional skewness and future Bitcoin returns is different depending on the sample period. In the first subsample (2018–2020), a period of relative calm in the Bitcoin market, the relationship is negative, which is in line with that found in the literature. However, in the second subsample (2021–2022), a period of major turmoil in the Bitcoin market, the relationship is positive, which is consistent with that found in previous papers on the relationship between conditional market skewness and future index returns during crisis periods. Based on these results, a dynamic buy and sell strategy of buying or selling Bitcoin based on the estimated conditional skewness is proposed. This dynamic strategy outperforms a static buy-and-hold strategy. The profitability of this strategy can be viewed as the reward that investors demand for bearing the risk associated with the changing conditions in the cryptocurrency market that generate time-varying expected returns. DO - 10.1016/J.QREF.2024.101868 UR - https://portalcientifico.uah.es/documentos/6665f285be265147d9b62f2e DP - Dialnet - Portal de la Investigación ER -