We develop metrics based on Shapley values for interpreting time-series forecasting models, including "black-box" models from machine learning. Our metrics are model agnostic, so that they are applicable to any model (linear or nonlinear, parametric or nonparametric). Two of the metrics,...

The concept of a forecast horizon for a given player with respect to the set of equilibria in an abstract dynamic game is defined. Consideration is then given to the case of a stochastic dynamic game with memory-less perfect state information and the Nash equilibrium concept. Sufficient...