Estimating of Bellman function via suboptimal strategies

The paper concerns approximate dynamic decision making. It deals with solving Bellman equation to obtain the Bellman function via so-called suboptimal strategies. The sub-optimal strategies are strategies reflecting the revision of the realized past decisions. The revision is conditioned on availability of future knowledge. Suboptimal strategies help to transform the estimation of Bellman function to solving system of algebraic equations, when parametrized form of Bellman function is used. The presented approach is applied to futures trading task.