​In this article we propose a dynamic factor framework for modeling and forecasting financial and commodity term structures in a unified global setting. The novelty of our approach is that it exploits a large set of information (i.e. data properties, time and forward dimensions, and cross-country, market, sector and weather dimensions) summarized in a set of heteroskedastic components that have a clear time series interpretation and that can be modeled dynamically to generate forecasts in real-time. The approach is motivated by evidence of rising financial integration, and interdependence between commodity and asset markets. We employ a battery of in-sample and out-of-sample techniques to evaluate our framework and concentrate on relevant statistical and economic performance measures. To preview our results with practical implications, we find that our framework provides significant in-sample information in terms of product specific factors and commonalities driving commodity and financial markets. Moreover, the specification proposed for modelling the dynamics of financial and commodity term structures generates accurate out-of-sample interval and point forecasts and leads to variance reduction when hedging a portfolio made up of spot and futures contracts.