A carregar...
Publicação
A deep learning algorithm to 0ptimize multi-asset portfolio returns based on macroeconomic variables - applying hierarchical risk parity
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.28 MB | Adobe PDF |
Orientador(es)
Resumo(s)
The individual investor does not usually take advantage of quantitative methods to invest, such
as Machine Learning algorithms. Additionally, the individual investor is often tempted to time
the market. This paper proposes an investment strategy that dynamically allocates weights
based on macroeconomic variables, aimed at facilitating the access to advanced methods for
the individual investor, diminish the influence of emotional behavior of the investor in his
personal investments, and corresponding attempts to time the market. It does so by studying
past relationships between macroeconomic variables and the optimal asset weights that would
have led to a good performance in the past. These relationships are examined by means of an
Artificial Neural Network. Finally, to better take advantage of the different economic scenarios,
and to potentially benefit from diversification gains, different asset classes were considered,
beyond the most adopted bonds and equity.
Moreover, this dissertation comprehends the study of different portfolio optimization methods.
Descrição
Palavras-chave
Hierarchical risk parity Machine learning Macro Portfolio optimization