(I wrote it in Matlab during my home trial period of 30 days.

Wow! I'm so glad to see somebody else using this info and improving upon it! I have a simulator as well. This means that in this situation, the optimal strategy is to ignore an immediate opportunity to lock and let the lockable row run for a bit. Starting player is rotatedĤ1.33 P3A(MinGapOffn:3) StartedFrom: R23456cĥ8.67 BMH1(Bmhowe34) StartedFrom: R4 Y34 Ga Bb Starting player is rotatedĦ0.36 P3A(MinGapOffn:3) StartedFrom: R23456ģ9.64 BMH1(Bmhowe34) StartedFrom: R4 Y34 Ga Bbīut in that same situation, immediately seizing an opportunity to lock red only gives P3A the win 41% of the time. When player P3A has 2,3,4,5,6 in the red row and BMH1 has red 4, yellow 3,4, green 10 and blue 11. I’ve also played around with some situations that show that it is not always optimal to lock as soon as you can. I also found that starting order matters – (non-rotated start results): Starting player is NOT rotated The difference is more pronounced with more players. I ran the two algorithms head-to-head and yours came out a clear winner. Otherwise all good moves have negative scores and would not be executed due to the no-play threshold of 0. Should be: –Transient to transient: ExpRolls(CurrentState) - ExpRolls(NextState) (I noticed that an error exists in the PDF: –Transient to transient: ExpRolls(NextState) – ExpRolls(CurrentState) Btw… I’ve always wondered if the sprint-to-the-lock was the best strategy or a more conservative approach was better. I also implemented your algorithm to see how it stood up to mine. Plus some other rules around when to take an X. Break any ties by taking the move that brings the most points. Unless the row is empty, then allow up to three, ie (2,3,4,5 to start empty row). If the same gaps exist for multiple moves pick the one that brings the most points. But if the row is empty, then allow up to three gaps, ie (2,3,4,5 to start empty row). Pick the move that minimizes gaps in the rows. I haven’t had time to look for an optimal strategy yet but did implement my standard experience-based algorithm that I use when playing the game. I just got it working then stumbled across your post. So I wrote a Qwixx game simulator to try and find some answers. It has always bugged me when making a tough decision in Qwixx knowing that an optimal solution existed but I did not know what that was. Excellent! Nice to see some theoretical work done on Qwixx strategy.

Qwixx rules pdf