摘要
The spirit of now in nowcasting suggests expanding the current to include the near future.Decision theory is then developed by incorporating the consequences of actions into the present.With the future falling into the present discounting it is no longer permitted.Value functions are then observed to be determinate only up to scale and shift that are then locked down by fixing values arbitrarily in two selected states,much like declaring water to freeze and boil at zero and a hundred degrees celsius.The locked down value functions associated policy functions are seen to exist in decision contexts in where the only time is now.Examples are studied in univariate and multivariate dimensions for the decision state space and the dimension of shocks delivering state transitions.The policy functions are expanded from realisitic training sets to the full state space using Gaussian Process Regression.They are implemented on real data with reported performances.
