Bayesian forecasting and portfolio decisions using dynamic dependent sparse factor models

We extend recently introduced latent threshold dynamic models to include dependencies among dynamic latent factors underlying multivariate volatility. With an ability to induce time-varying sparsity into factor loadings, these models now also allow time-varying correlations among factors; this may be exploited to improve volatility forecasts. We couple multi-period, out-of-sample forecasting with portfolio analysis using standard and novel benchmark neutral portfolios. Detailed studies of stock index and FX time series include: multi-period, out-of-sample forecasting, statistical model comparisons, portfolio performance using raw returns, risk-adjusted returns and portfolio volatility. We find uniform improvements on all measures relative to standard dynamic factor models. This is due to the parsimony of latent threshold models and their ability to exploit between-factor correlations to improve volatility characterization and prediction. These advances will interest financial analysts, investors and practitioners, as well as modeling researchers.