Quote Originally Posted by Gaius Scribonius Curio
Carrhae, however was a rarity at the time. The Pahlavi Surenas' use of the camel train was innovative for the time. The constant resupplying led to a huge number of arrows being fired at the romans, and despite their discipline and training there was nothing they could do against this hail of arrows. Publius Crassus led the few units of gallic cavalry to drive off the archers and was pounced on by a superior force of cataphracts I believe. Which left the Romans without a way to drive off the horse archers.
The point is that the HI though absolutely superior force in Shock close-combat, could not inflict any damage, because their missiles pila, and the javellins were out-ranged and outclassed by the bow. Marching in any direction to attack appeared pointless as the HA simply retreated to a distance and then re-engaged.

The "few units" were 4,000 Cavalry detachment trying to break out of the encirclement. The Cataphracts were essential for repulsing break out forces of the encirclement. But for the organised breakout by Roman horse they waited to attack when they were absolutely isolated from the main Roman force. Mauling the Roman Cavalry, allowed them also to be encircled and destoyed slowly by the light missile men.
AFAIK that was the first battle of Carrhae, during the night the Romans retreated to the samll fort at Carrhae itself, and then opted to withdraw to Syria, and were betrayed by a Parthian spy.
Crassus took the bad advice of a betraying ally, who encouraged a swift attack, and took a short cut through the desert plain, rather than keep to the river against the advice of his experienced commanders, where the army would stay in touch with the supply lines. Initially the skirmishers retreated and fell back, drawing the inevitably "victorious" overwhelming army on and deeper into peril. The battle was decisive, Crassus's son was killed leading the breakout attempt, and Crassus's army was effectively destroyed entirely.

Whilst it may not be a typical battle, it's a good example of why bad strategy by over confident incompetent commanders, who thought "Legions good, I be victorious" and undertook an unauthorised campaign for their own selfish political reasons.
As far as I'm concerned it was becoming far too easy, and as such I considered it an exploit, because the stubborn Scipii just kept retraining new legions to be massacred by my impossibly small force led by my ten star expert cavalry commander, expert defender, good attacker, uber general (who started off with one star).
Exactly... The Romans have to use money to purchase more suitable cavalry forces, and re-balance their army, so it has the right combined arms; or fallback to ground more suitable for it's forces. If the AI fails to do that, then it's a playing strength issue.
On your other point though, I have noticed that given enough time the Ai will adapt, (above I was just getting bored), The Seleucid Empire at first, didn't rate me and sent small force of varying quality, which failed, so they began sending large force consisting of phalanxes, which again failed. In the final stages of the war (actually the fourth in a long series of them after they betrayed our alliance against Egypt), they began to use lots of chariots (which to begin with were actually really effective against me, I soon learnt)
Interesting. Against Barbs, the settlement loss rate is so fast it's hard to see that effect. When I suspected it, I may have been mopping up "old" stacks, produced earlier in the game, which were unbalanced with spammed Phalanx spearmen.
withdrawing when out of missiles, and reattacking next turn. It is a viable strategy, but a long one. And I like the idea of an 'indecisive' result, makes it more fair, the ostensible victor would have an untenable position and have to retreat like the loser, perhaps?
Need space to fall back. Of course the AI probably wouldn't realise it needed to fall back, on phyrric victories. But would leave it's mauled army exposed to a counter-attack.

What it really would do is set expansion limits, as it becomes economically unviable