This thesis examines the possibility of predicting the equity risk premium using an advanced machine learning model. The objective of this thesis was to develop a machine learning-based prediction model and estimate the implied risk premium of the US capital market based on Professor Damodaran's FCFE model (2024) over a one-year horizon. The final prediction is compared with the actual implied risk premium published by Damodaran (2024) and then with the prediction using the ARIMA model and ... zobrazit celý abstraktThis thesis examines the possibility of predicting the equity risk premium using an advanced machine learning model. The objective of this thesis was to develop a machine learning-based prediction model and estimate the implied risk premium of the US capital market based on Professor Damodaran's FCFE model (2024) over a one-year horizon. The final prediction is compared with the actual implied risk premium published by Damodaran (2024) and then with the prediction using the ARIMA model and the historical risk premium, which is often considered to be the best estimate of future developments. The results of this thesis show that a suitably configured machine learning model provides a better estimate of the future equity risk premium compared to both the historical risk premium and the ARIMA model. The price for this accuracy is the low explainability of this prediction compared to the ARIMA model. Finally, the thesis is devoted to summarizing the results obtained and formulating conclusions about the real-world applicability of the developed model. |