Are you sure about that? Let’s see an example. I assume that with troops and emblems, Odin has an attack stat of 1000. With BK buff that makes it 1450 (assuming minimum 45%, but could be way more depending on whether he was hit with tiles prior to his special going off). And I can see 5 holy tiles on the board. That makes his damage 100 + 18x5 = 190%. Now let’s assume that he hits a hero with a standard defense stat of 800. According to the damage formula,
Damage ≈ 100 x ( θ x Att / Def ) ^ 1.35
Att = 1000 x 1.45 x 1.9 x 1.2
Def = 800
θ varies from 0.606 to 1.649
If we now plugin all the values, the damage can vary from 345 - 1333.
As we can see, that’s a huge amount of variation (about 1000 HP). But both of these are extreme cases. Most average cases will likely lie somewhere in between. All the values you see in the picture are within this theoretical range, so how can we claim that it is not due to the random parameter? 
Once again, there is no definitive proof as we just investigated. That’s why I said in my first post, that we would need to study many more such cases, and if the pattern repeats itself, then only we can definitely zero down on a more definitive answer. Until then, we have to chalk it up to the random paramter for now.