A carregar...
Publicação
Forecasting stock index volatility - a comparison of models
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 585.35 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
This thesis explores the useof popularmachine learning algorithms(K-Nearest NeighborandRandom Forest)and compares them to traditional techniques (Random Walk, ARIMAand GARCH) for forecastingone-day, one-week, one-monthand one-quarter volatilityusingThe OsloStock ExchangeAll Share Index. A number of error metrics are applied(RMSE, MAE, MAPE and R-squared)in order to compare their results.Machine learning methods are shown to forecast thechanges in volatilityto some extent, however, evidence isfound favouringtheARIMAmodel when forecastingvolatility time series.
Descrição
Palavras-chave
Volatility Forecasting Garch Machine learning