I hope no one minds if I repost my AI assessment from the other thread, experience won in my two independence campaigns, one Britain, one US. It includes my strategies to exploit the AI's mistakes and gain major victories in both.
I played the whole British one with 1.00 and most of the US too, with 1.01 the AI was improved, at least it managed to defend Wilmington though I had sent a strong army, and it kept a strong garrison in Camden, which was prudent. Before 1.01 it had actually achieved nothing on the strategic level, except through scripted events of course.
1.) Most important: The AI does not rest it's troops like the players!! The player will usually follow the maxim:
MOVE -> REST -> ATTACK X -> REST in X
AI units I always found to be under strength and could thereby be disposed of!
2.) Retreat into town with supply > 15 to rest: obvious
3.) Retreat into town with supply > 15 when winter approaches and stay there until weather improves, ideally major cities like Boston, New York, Charleston, Savannah etc
4.) Concentrate on the strategically really important objectives:
Ft Ticonderoga, Albany -> as a required bases for any strategy in the Northwest
Boston, New York -> have extreme supply level to sustain large armies and the harbours that can repair huge fleets
Richmond, Norfolk -> to control the complete road / naval transport between North and South. This alone can completely ruin the AI.
Charleston, Savannah -> the only real supply bases in the south
there could of course be differing opinions on nthis ..
5.) If you have the major objectives, keep them!! The Ai most of the times makes the mistake of offering me hard won objectives by either leaving no garrison or only a single unit. Every offensive will fail if I can take Boston or Montreal behind their backs! This is how I annihilated the british expedition to Boston in only some months - when I took Boston they were stranded!!
6.) keep single unit garrisons in every town along the "front" and if possible in some depth behind the front. It's surprising how often the AI sieges and then achieves nothing because I left only 1 militia unit there. I will then often have taken their base so they will be lost in 1-2 turns. So the 1 unit defensive strategy pays off.
7.) Use the assault command before you retreat. Any broken off siege is a big setback (because if you follow the principle you will have to REST), so it is usually worth the risk to assault (I took many towns that way). EDIT: of course only do this when the garrison is weak:
8.) Stack optimization: obvious, maybe the player will alway have the upper hand here .. there is too much that can be done here, like fast moving stacks with only cavalry and light inf, followed by main army stacks with artillery and supply, and such.
And yeah did I say: [SIZE="3"]REST[/size] ? I only ever attack with fully rested armies, I will move the shaken ones into another stack and use them as garrison.
Thats all I can think of now, these are all unforgiveable mistakes that experienced players will punish severely.
If anyones interested, savegame can be downloaded [color="DarkRed"]here[/color]. The action is now completely concentrated to Carolina between Camden and Wilmington, where the AI holds it's last straw with Cornwallis. But as I already relocated a lot of armies to the south the game is practically over.
I used 7z because it compresses age saves 20 times better than zip! No joke.