This is the "time horizon" problem. There's more theory about this than you ever care to know in books about reinforcement learning. A longer time horizon is b

This is the "time horizon" problem. There's more theory about this than you ever care to know in books about reinforcement learning. A longer time horizon is better, but you'll need to learn that at a shorter horizon of course.