Regret-Optimized Portfolio Enhancement through Deep Reinforcement Learning and Future Looking Rewards