In the present dissertation, I apply the Barroso and Saxena (2022) shrinkage methodology –also known as the Galton – to portfolio optimization’s inputs in the Chinese market (for the OOS period from January of 2011 until December of 2021). Each portfolio is a selection of the 50 biggest stocks in terms of market capitalization, and weights are rebalanced monthly. The Galton is based on the incorporation of OOS errors into input estimation to overcome the incapability of historical data to reflect the tendency of mean regression that mean returns, variances, correlations and covariances have. I show that the Galton correction’s superiority in risk prediction holds for China, even though the baseline method achieves an average negative Sharpe ratio. Nonetheless, when microcaps are excluded, the method produces significantly high Sharpe ratios, being this method the best among all the analyzed optimized strategies in a MV scenario. When the number of portfolio constituents is decreased, however, results for the regular form of the Galton seem to become positive; the changes in estimation windows, in their turn, have a favorable effect on the version where microcaps are excluded, making the Sharpe ratio rise above one.
Date of Award | 25 Jan 2023 |
---|
Original language | English |
---|
Awarding Institution | - Universidade Católica Portuguesa
|
---|
Supervisor | Pedro Barroso (Supervisor) |
---|
- Asset allocation
- Galton
- China
- Shrinkage
- Regression to the mean
- Fama and MacBeth (1973) regressions
- Microcaps
The Galton correction in China: when forecasting learns from the OOS
Roque, M. L. D. S. C. (Student). 25 Jan 2023
Student thesis: Master's Thesis