Cyclops don't attack everyone, that's true. However, with 3 large units (and often 4, because during the training of Paladins/Champions, they can be split into 2 stacks) in a Knight's army, it's very difficult to position troops so that several stacks aren't hit at once. And Griffons can easily be caught in the splash damage.
If we're talking about Axel, it's obvious that a unit of what level will be in dangerous numbers, therefore, a level 5 unit won't pose a special threat.
Therefore, I see the main task of the Orc as destroying the Knight's trained uber-stack, while capturing everyone possible.

Let's go into more detail here. :) Since we're talking about training, how much more expensive is it to have 10 Paladins compared to 4 Paladins and 6 Inquisitors? The amount isn't that important (it's something around 9000, I think, if with normal skills). That is, the presence of any secondary skill solves the economic problem. This means that for each Cyclops, there can be 5(!) Paladins. Let's say there are 150 necessary Chieftains, which means there are approximately 30 Cyclops, and 150 Paladins under Axel (in our model calculation). We position the horses so that, if desired, they can attack, they will respond to the one-eyed creatures (if, of course, there is no silence). Taking into account the distribution of parameters for the Orc and the Knight, and the fact that there are more artifacts for Attack than for Defense, let's assume that the Orc's Attack is 5 higher than the Knight's Defense, and the Orc's Defense and the Knight's Attack are equal. Now let's calculate the damage of the Cyclops during the first attack on the Paladins:
30 * 51 * 1.15 * 0.7 * 1.3 * 1.55 = 2482, i.e., approximately 25 Paladins minus (1.3 - the averaged model multiplier from Expert Luck)
And in response:
0.3 * 25 * 125 * 1.15 / 1.15 = 937.5, i.e., approximately 4 Cyclops minus
0.7 * 25 * 125 * (1.15 / 1.15) * 1100 / 6750 = 356.5 rage minus (i.e., with a decent probability, the rage is already less than zero)
It's not particularly great for the Orc... :rolleyes:
Now about the effectiveness of Crushing Blow. There is no strict description in the manual (maybe I had an old version?), but I assume that it is identical to Stun, but only in an area, i.e.
0.25 - 0.03(15000/6750) = 0.1834, which means that the probability of stunning, even with Soldier's Luck, is approximately 1/3.
I don't think it makes sense to accurately calculate the second attack of the Cyclops on the Paladins before the hero's turn; it's quite obvious that more than a hundred of them will remain alive, i.e., less than a third (!) of the stack will be killed. It doesn't really seem like a rout in the Knight's army. It turns out that even by focusing only on the uber-stack, it won't be possible to quickly remove it from the battle. Next, depending on the situation: either Subjugation on the Cyclops, or Vampirism on the Paladins, and game over for the Orc. Only resistance and a good roll will save him - Subjugation will not work, Vampirism will be cast too late or too early (silence, roar, etc.)
The main question is - how is it with training? Because, as the calculation shows, even a simple increase in monks can, with a decent probability, become an insurmountable obstacle for the Orc. Of course, if the focus is on 1 stack, then Darkness is our choice. Otherwise, Roar + Empathy, and it's unclear where the money went.

They won't return, almost guaranteed. Moreover, the Knight may not even have time to cast Vampirism, because in this case, he may miss the opportunity to control the Cyclops. The Knight's spell power, taking into account the reduction from the Orc's Spell Power perk, should be enough for at least 2 turns, so that the Subjugation-Vampirism-Subjugation combo works; otherwise, you can forget about Vampirism and just recast Subjugation.
Yes, with Readiness and Vampirism, anchors (both the 1st and 2nd options) can be excellent. However, the Knight is not a caster. He is not Orson, who relies on a Zombie anchor and defeats the enemy with magic, and he is not a Warlock with anchors of Minotaurs and Hydras, pouring Chaos on the enemy. Who will attack? Well, these 2 anchors will remain, which the Orc will not touch. He will cut down the rest of the troops, run away when he feels that "it's enough," and return with the remaining unused army.

It's clear that the "anchor" + Darkness tactic was honestly stolen from the Necropolis. And no one can apply it better than the undead, but if the artifacts are not for Attack/Defense, but for Magic (don't forget about +5 for Darkness), then it may turn out to be viable for the Knight. Of course, the slow (and therefore "anchor") Cyclops will attack, and not pure magic.

There is no point in commenting on the replay from Lordesat. In a rush, 2 "adjustments" MUST give a turn (after morale - one). So, 30 Chieftains in a stack is not an option.:smile38: