How to play well in a sum of hot games has first been studied by Yedwab. Cazenave did an empirical study on sums of 4-stop games of the form a|b||c|d. In the paper below, we look at sums of more general games, and show that even a shallow search can greatly improve upon static methods such as hotstrat. We have a few more unpublished results in this area.
In 2006, Andraos et al. found an interesting characterization of combinatorial games and how the different static methods behave on those. They also created new improved algorithms that utilize their characterization.
Links:
Smart search algorithms for sums of hot games is still a largely unexplored field. It has potentially very big impact on creating a strong endgame player for Go.
M. Müller and Z. Li. Locally informed global search for sums of combinatorial games. In J. van den Herik, Y. Björnsson, and N. Netanyahu, editors, Computers and Games: 4th International Conference, CG 2004, volume 3846 of Lecture Notes in Computer Science, pages 273-284, Ramat-Gan, Israel, 2006. Springer.
Martin Müller