That works for me :)
Working with the AI is not the best solution, but at least it's fairly consistent. I can run it over and over again on the same map and it never learns a thing
I'm testing with a 5:2 ratio on a 100 hex front. If the defender has 2000 points, it loses 67% of the time, but if it has 6000 points, it wins 83% of the time. The attacker has 5000 and 15000 points respectively, so the 5:2 ratio is the same. My guess here is the extra defensive points allows the defender more mines and the attacking AI doesn't handle mines well, if you don't count finding them with Panthers as "well".
As mentioned by Vesku, the point ratio would do well with some base and then other factors would have to be considered. Ratio alone is not enough to cover all situations. Taking player skill out of the picture, you still have to consider total defensive force verses area that has to be defended, length of battle, visibility and probably a few other factors. The player skill impacts what force they select and how they use it, but should be why someone wins or loses.