Evaluating dynamic covariance matrix forecasting and portfolio optimization
MetadataShow full item record
- Master Thesis 
In this thesis we have evaluated the covariance forecasting ability of the simple moving average, the exponential moving average and the dynamic conditional correlation models. Overall we found that a dynamic portfolio can gain significant improvements by implementing a multivariate GARCH forecast. We further divided the global investment universe into sectors and regions in order to investigate the relative portfolio performance of several asset allocation strategies with both variance and conditional value at risk as a risk measure. We found that the choice of risk measure does not seem to heavily impact the asset allocation. As comparison to the dynamic portfolios we added regional/sector portfolios which where rebalanced after a 3% threshold rule. The regional portfolio was constructed to mimic the current strategy of the Norwegian Pension Fund Global. The max Sharpe portfolio for regions had the highest risk adjusted return, but suffered from a very high turnover. After being modified however, this strategy turned out to be superior even after transaction costs were imposed.