This thesis tests the predictive power of the stock market. It suggests new ways on how to improve it is predictive ability. Four main methods were used: combining individual predictors, assigning different weights to the best individual forecasts, modelling predictive ability and explanatory power. I show that combining the individual forecasts into dynamic models yields better results when compared with the models in Rapach et al. (2013). I find that the best model was highly dependent on the frequency and the geographical location. Lastly, the economic significance of using predictive regressions in Asset Allocation was also tested. I conclude that statistical improvements more often than not result in artificial enhancements of the predictive power, which do not translate into practical economic gains.
| Date of Award | 27 Apr 2021 |
|---|
| Original language | English |
|---|
| Awarding Institution | - Universidade Católica Portuguesa
|
|---|
| Supervisor | José Faias (Supervisor) |
|---|
- Out-of-sample regression
- Rapach et al.
- Value weighted mean
- Asset allocation
- Predictor analysis
Predictability of stock market returns and it is importance in Asset Allocation in the US
Gomes, T. F. A. (Student). 27 Apr 2021
Student thesis: Master's Thesis