The basic strategy for war is to tech until you get a significant military advantage:
There are multiple points in the game where an advantage is less meaningful, though of course if you can get one go for it.
Eg:
- Catapults when your opponent has none
Catapults are barely a threat from the AI should it have them. No reason to stop a war upon seeing cats or to avoid building up because you think the opponent will get them. Longbows and especially castles/protective are the anti-cats.
- War elephents when your opponent doesn't yet have longbows
Capable of easily taking combat II for 9.6 base str, even marginal collateral puts longbows behind elephants. This troop combo doesn't die until engineering, which walls the cat bombardment and offers a strong counter unit, with more problems around the corner.
- Curaissers before any opponent has pikemen
Pikes win hammer for hammer, but curis win straight up and have excellent mobility. Realistically curis are pretty good until they start pushing rifles.
- Rifles / Cannon before an opponent has them
Cannons obviously prefer fighting pre-rifles, but CR II cannon vs rifle isn't exactly murderer's row. Try to deny horse though or bring more rifles.
- Unique unit that is game changing in its time like immortals, praetorians, samurai etc
War chariots are superior offensive units to immortals, but both are quite good. Quechas, Janissaries, Oromos, Keshiks, Cho Ko Nus, Cataphracts, Conquistadors, and vultures are all worth a strong look too.
Then go all out - make nothing but units in every city. Chop them, whip them, draft them. Then throw them en-masse at the enemy to destroy the enemy as quickly as possible. Successful wars are not finessed or elegant - they are brutal with high casualties but quick. If you are in a slow war of attrition then you didn't bring a big enough army - sue for peace and repeat when you are ready.
This is one way to play it and viable. It is not the only way...not on any difficulty. Building up units over time with the expectation of mass upgrade is perfectly viable. City specialization where several cities do not build units is fine. A slow, 30-40 turn medieval slog is fine too as long as you deal critical damage within the first 10-15 turns such that only 1 or at most 2 cities can easily handle your side of the attrition (doing this allows you 5:1 kills/deaths and easy cities and the window before rifles is large). Later in the game, brute force things like infantry/arty (a personal favorite) competes with excessively powerful finesse moves like amphibious attack + fighter support, or the brutal but tactical use of nuclear arms.
If you are going for conquest then brute force wins. No finesse. A really big stack (or stacks later) that can absorb anything the enemy can throw at it and which has enough seige to take down the defenses of the time quickly.
This ignores 100% mounted charges which combine some elements of brute force and finesse. Doing that forces you to move tactically and take care when - and how - you engage opposing forces, but the result can be a fast war with minimal chance for the AI to stop it.
The army will have some stack defenders and some specialist city raiders, but generally it doesn't matter too much - size is key. A big stack can just shrug off the enemy seige attacks (its worth including even obsolete units just to bulk up your stacks - they are still good for mopups).
Don't get me wrong though. I understand the value of stack bulk quite well and also advocate using older units for mopping up, holding to use city capture gold for upgrades (or tech sale/extortion), or as post-conquest vassal garrison to keep city revolt chance at 0% while you build a shinier army to continue the beatdown on other civs. However, profitable parity war is doable at most points in the game!
1. Ancient - massed axe/spear (more so axes) using chops to focus hammers to temporarily trump AI bonuses and non-military focus early on. Chariots are arguably even better than axes when rushing.
2. Classical - if the target is not protective, there really aren't many pre-longbow defenders that do very well vs catapults, so losses won't be bad. Getting enough #'s here is hard though. Elephants get a gold star here - nothing safely attacks elephant + axe with winning odds until muskets.
3. Medieval - Tricky. Resource denial starts to matter the most here until the era of oil and uranium. Deny horse or iron and you don't see knights. Use longbows + a hill city and the AI will likely suicide its entire offensive stack while you DEFEND for a change, and with a heavy advantage in doing so. Without knights, not much can effectively flank or defend against trebs for long. Slow as these are, continuing with infra mid-war isn't hard and you have 50+ turns from declaration until the AI gets anything worrisome, probably more since you *will* cripple its tech. In a recent game I attacked spain in this era and did better than a lot of other players because I denied metal and killed units before they upgraded...notably choking off iron = no UU.
4. Renaissance - Tried and true. Cannon, cavalry, and rifles all work. Cuirassers are popular too. Take your pick. In tech parity, however, only rifle + cannons shine. But boy, do they shine. Airships can help keep cannon loss down btw.
5. Industrial - Infantry and arty are solid attackers and difficult to counter before, say, advance flight (!). Nothing defends particularly well vs CR II arty. Not tanks, not infantry, definitely not machine guns (they ignore collateral but are terrible straight-up). Big window for minimal losses even at parity.
6. Modern - Tanks + bomber time! Or nuke/para. Or marines + fighters. Or some other combo. War at tech parity in this era is very, very easy vs the AI and won't feel much like parity at all.
7. Future - Very similar to modern...any semblance of tactics will leave the AI helpless unless you are vastly undersized relative to it.