9#%002M

There are two extreme scenarios: either early contact or a prolonged stalemate. Treating one exacerbates the other. That's why testing in real games is necessary to find the optimal balance. In the current version, I've set the guard to 6,000, considering that breaking through two such defenses without significant losses is difficult but possible. This means the map isn't rigidly divided into zones in the first week, but allows for rushes, although these are infrequent and situational, depending on the map layout. If this isn't required, the guard can be increased to 9,000, which would delay contact until the second week. Regarding road speed, I think it should be set so that it's not possible to reach secondary locations from the main castle in one turn (in most cases). Otherwise, it would be possible to capture multiple castles in a single turn by buying a hero in one castle and having them reach the next castle. This increases the influence of a particular player. Visually, the medium road speeds seem to satisfy this condition, but it needs to be verified. I haven't come up with a specific criterion for resources; I'm not sure what the standard building cost is on a Medium difficulty map.