I've been working on a tool to help me design monsters, specifically doing CR calculations as described in the DMG. In order to test it, I've tried it out on some monsters in the Monster Manual, which of course are giving me different results.
I just want to say up front that I'm fully aware that the DMG method doesn't align with what's actually in the MM, and that a different calculation was used for it. And for my purposes that's ok, but it does make testing it rather difficult. I also understand that playtesting can influence the CR.
So what I want to check is: For these 3 test monsters, is my calculation correct, or have I overlooked something? And if it is correct, is there any widely accepted or documented reason for these monsters specifically to have different CRs?
Wolf (MM CR: 1/4, calculated CR: 1/2)
Damage per round: 7
Attack bonus: 4 + 1 (for Pack Tactics) = 5
Defensive CR: 1/8
Offensive CR: 1
Average CR: 0.625 rounded = 1/2
Wight (MM CR: 3, calculated CR: 2)
HP: 45 x 2 (for damage resistances) = 90
Damage per round: 2 x 6 (longsword) = 12
Attack bonus: 4
Defensive CR: 2
Offensive CR: 1
Average CR: 1.5 rounded = 2
Planetar (MM CR: 16, calculated CR: 15)
HP: 200 x 1.25 (for damage resistances) = 250
AC: 19 + 2 (magic resistance) + 2 (3 saving throws) = 23
Damage per round: 2 x 43 (angelic greatsword) = 86
Attack bonus: 12
Defensive CR: 12 + 3 (adjust for AC) = 15
Offensive CR: 13 + 2 (adjust for attack) = 15
Average CR: 15
The advice related to challenge ratings in this article can be used to explain the Planetar and Wight. Both of those monsters can drop a PC in one round, so refining and applying the approach referred to for Ogres (i.e. increase it's CR by one if it can drop a PC in one round) results is CRs of 3 and 16 respectively. Statistically there are a few things that don't make sense with that approach, but then again there is quite a lot about the CR system that is statistically questionable.
None of that explains the wolf though.