Interesting script. I guess it still ends pretty similar. Seems like the problem with tuning pure BM vs BM is that there are so few non-obvious decisions, where one script gets an opportunity to really prove itself. So, regardless of whether you use turns, deck position, or money, you end up calibrating the thresholds to get provinces after the third reshuffle and going hard green when provinces run low. Then the only opportunity to manage your luck is for a couple turns in the midgame where the decisions are very complicated and one variable probably can't tell you what to do. And even then luck will frequently still get the better of you.
I think a resilient script would be really cool but I can't see how you would make a solid one based on any single one of the metrics available. Unfortunately, the simulator code isn't really built for multivariable approaches (i.e. take the total money in your deck divided by the cards and do the same for your opponent and add the result and...ad infinitum). Maybe at some point there will be user-scripted conditions?
The Cheese script actually has a modified PPR and it took me a while to track down what this does. It ducks the penultimate province when 1 or 2 points behind even when there are no duchies in the supply. This performs badly against my turns + PPR script but is actually an advantage against the "betterer" script! I think there are some similar triangular issues with late copper purchasing if someone wants to investigate them, however the variance of the results makes my PC too slow to run enough tests.
Haha, didn't even notice that. Whoops. I don't think it is totally surprising that there is a bit rock-paper-scissors going on there, especially when we are talking about a different of small percentages. I don't think it means something significant, but rather is a byproduct of the fact that each script is so simple that it handles certain rare situations extremely poorly. If you make a move that inadvertently pushes another script into one of its degenerate situations you increase your chance to win by "confusing" it.