Loki's time management only consists of dividing the time left (given by for example "btime/wtime") left with 30 or the amount of moves to go if specified.
This results in Loki burning a lot of clock in the opening and playing poorly in the endgame as a results of only having seconds left.
New time control logic is planned to firstly be implemented in Loki 5.0. The implementation should take into account:
The game-phase. It is better to use a lot of time in the middlegame in order to get a big advantage, and then using less in the endgame. This should be possible considering that a good position is easily won in the endgame.
The complexity (a term that needs to be added). This will be used to give more time for seemingly harder positions and less for easier ones.
While searching, two things should be implemented: 1) Easy-move recognition for stopping the search if one move seems clearly superior to all others, and 2) Panic time. If the position from one iteration to another suddenly seems a lot worse, we need to allocate extra time to try to solve the issue.
Loki's time management only consists of dividing the time left (given by for example "btime/wtime") left with 30 or the amount of moves to go if specified. This results in Loki burning a lot of clock in the opening and playing poorly in the endgame as a results of only having seconds left.
This should be improved upon.