Prediction: Attack chance of success

The to do plan for my game now looks like the following:

A. Run a prediction of what the enemy will do short term.

B. Run a prediction for what I will do short term. This is useful because I can see in advance if my attack will have success. If my prediction attack fails (entrance to the enemy base is blocked, units die before they reach the enemy base due to enemy tank fire when tanks are on high ground and can not be reached etc. ) I can chose a better course of action, another version of attack that is, I can attack an expansion instead or I can chose another route to the enemy base.

What the enemy will do short term usually consists of two or three things: Build a building, build units from that building (about 4 soldiers) attack with newly built units. The first wave of attack (something that we can build in 5 minutes) consists of one building and 4 soldiers. The second wave of attack consists of two buildings (one that already exist) 4 soldiers and 2 tanks (again something we can build in 5 minutes). The third wave of attack will be more buildings and units but again something we can build in about 5 minutes of gameplay.

[edit]

Recently I've had the following idea: When you start a game you have two or three options. Each option is a distinct course of action. Basically you're weighting two or three tactics.
For each option imagine two ways the enemy will react to that course of action. For each of those two reactions figure out a counter or maybe two ways of counter.
Finally chose the option with the best perspectives. Basically you're figuring out from the start all the possible moves.