Towards optimal doubly robust estimation of heterogeneous causal effects | Steady Practice | SteadyPractice