Strategic farsighted learning in competitive multi-agent games