Heroes 5.5 WarGame Edition - Tactical Edition.

Um... I'm dividing this into 2 parts:
1. Are additional perks of the "strategic" skill (+units, +money for units) considered to be "working in battle"? For example, like the Order racial one (the part that handles retraining) - does it count?
2. Logistics +10/+20/+30 is very-very-very chunky. That's basically a level 30 Wingael, who gets nerfed and banned in every possible mod.

If the skill can be naturally integrated into the arena, it counts. For example, necromancy or army gathering (plus one more meat).
The Arena is a map created specifically for fast leveling and battles with an opponent.
But specifically, money from a skill isn't an interesting idea for such an arena, because additional levels are bought with that same money; it ends up being simple level compensation.

As for Logistics - I don't think it's chunky compared to Attack or Defense at 10/20/30 (moreover, it's incorrect to compare 3 skill levels to some special ability; of course a triple skill will be stronger).
Why is Wingael nerfed? Because in the original, elves always reach and hit on the very first turn. It's the most disgusting kind of rush. That isn't the case here; the advantage in starting tempo is softer. Furthermore, this ability is tailored for low-initiative units primarily, allowing their first-turn advance to be significantly accelerated.

Added after 1 hour 1 minute

More. For example, there are unusable perks (Archer's Rise, for instance). Or a perk whose entire effect is simply granting an active ability (Oversight, Reinforcements). These can be given as a spell, while the perk itself is used for other purposes. You could easily make a skill out of them, where the effect is achieved via scripts and/or buffs.

That's clear; the tools are there. The question is about specific interesting ideas.

Just stats - very boring.

But natural and in demand, for lack of anything better (for now).

Is this the one that glitches at Mentor's and doesn't reset stats? :D

You can actually disable granting stats for education in the exe and grant them via scripts however you want (or not grant them at all).

I also really liked and still like it. The advanced class system, by the way, is very interesting.

Yes, it's interesting; I skimmed through them - but the bonuses aren't interesting. Everything, as usual, comes down to interesting bonuses.

This is where the unfairness always annoyed me. Basic Luck, which a hero spends a skill point on, gives banal +1 Luck, which is perfectly replaced by a minor clover leaf in terms of effect. Same for Leadership. I would have added additional perks, considering the possibility of using combat scripts.

That's exactly what I did for luck - plus one random parameter per level. But it's still a debatable decision (I didn't do it for leadership since those abilities are already strong). Although, as recently discovered - retribution isn't that powerful; only direct damage is boosted (it doesn't work on counter-attacks).

I disagree. Yes, the hero is not a unit, but can take an active part in battle with their attacks. There are plenty of options here - shifting forward on the ATB after using a basic attack, giving a chance for an automatic second attack on the same target, increasing the hero's level during a direct attack (if possible), improved Oversight/Cursed Signs/Reinforcements.

Unfortunately, combat script capabilities are still weak.
You can track the start of a turn, but not its end, and even more so, you cannot determine which action was performed.

This is exactly how it's implemented via governor scripts in MMH5.5. Why not? More details slightly below.

I haven't tried what that looks like, so I have doubts. Why is this needed for a combat hero? Why is this needed on the arena? What specific interest do such skills bring to battle?

Is it necessary? Yes. It's variability. Either you dump all available experience from neutrals into one hero. You get, say, level N and a good killer-hero. Or you spend experience on a killer-hero and a support-hero, who brings extra units and extra cash for those extra units. In the end, the killer-hero will be M levels lower, but with an additional army. Who wins? Quality or quantity? That's how you build your character.

Everything is fine, except for the leveling part. For a combat hero, these are dead abilities. Accordingly, the same goals (leveling several heroes) are better achieved by other means.

Strange choice for orcs - wolves and Manticoras. Melee fighters without access to Blood Rage.

I haven't looked into NCF yet.

That's true. I would actually set up a dialogue with the MMH5.5 creators, so that truth could be born from debate.

I've already spoken with them, put forward my thoughts on skills and spells - but they have their own position because they play their own game. For example, combat abilities cannot be strengthened because combatants would start tearing mages apart (as if it weren't already the case with huge stacks of meat!). Buffs cannot be strengthened for the same reason. And so on.

Here are my thoughts on this. Leveling out losses when breaking through = leveling out "hardcore players" and "casual players".

I was one of those hardcore players who took utopias with furies and crystals, or broke through with t6 crossbowmen in the first week. But it gets tiring; endless abuses that are so effective they leave no room for other options.
The simplest example is shooters. All else being equal in PvE, preference will always be given to shooters because they are a key unit whose property, when dominant, reduces losses to a minimum due to non-contact combat. Therefore, the balance of PvP and PvE differs greatly. And so, to smooth out this difference, it's logical to restore part of the losses (not all, but only 50%).
This idea won't be implemented anytime soon anyway. It's currently just in the idea stage.

Even now I struggle to perceive things like "mathematical expectation from Luck +5". I simply accept it as "+50% damage over a long period" (in the classic version, naturally). Valery Ivanovich tried to hammer his probability theory into my cortex. And in general, 60% of the brain is just fat.

Interestingly, in the case of such a discrete value as luck (it triggers or it doesn't), the geometric mean more accurately reflects the bonus efficiency. I modeled a clash between 2 stacks in Excel to test certain hypotheses and found that a 50% chance for 150% damage gives the same advantages as a stable +22.5% damage (not 25).
The same applies to damage ranges; stable damage is slightly stronger than mathematical expectation (but it's expressed much less clearly there because the damage distribution isn't discrete, unlike in the case of luck).

Added after 1 hour 31 minutes
I wrote that in MMH5.5 you cannot effectively convert spellcasting into physical stats through buffs. Well, it turns out I was wrong.
What are we talking about? Spells like stone skin, punishing, weakness, which work according to the formula: (Spellcasting/3) + 6 (bonus), (Spellcasting/10) + 1 (duration). Mass casts there push back by 0.15 ATB (meaning the hero's turn lasts 0.85 rounds). Now, specifically taking duration into account, spellcasting converts normally into physical stats (provided the hero has a sufficient number of mass buffs).
That is, from a balance perspective, their formula works.
Nevertheless, from a gameplay position, I don't consider such an approach successful. Because for most of the game your buffs will be short until you push spellcasting very high. And short buffs always mean playing a tactical rush, because it's impossible to create sufficient potential for troop enhancement (under multiple buffs) for non-rush play. You can't talk about having set up some combo if every 2 turns your hero does nothing but refresh the same buffs.
In that game, there are basically no prospects for short-term play (by their standards) for a buffer (while spellcasting is below 10-15), because they won't have time to wind up their magical potential before the buffs expire.

For this very reason, I completely tied buff duration to the skill; if you want to play with a large number of simultaneous buffs - max out the school skill; if you want to play with the maximum variety of buffs (Light+Dark), or buffs + something else (or a second magic school) - focus on long-lasting buffs that last twice as long and almost don't need refreshing. If you want maximum buff strength - focus on tier 2-3 buffs and pump spellcasting for them to the ceiling.
In WGE, blessing, evasion, weakness, and slow last 2/4/6/8 rounds (depending on skill), while curse, confusion, haste, skin, and might last 1/2/3/4 rounds.
Moreover, t1 spells (blessing, evasion, curse, confusion) affect everyone (but aren't boosted by spellcasting), while the others have a 4x4 area.
Skin, Might, Weakness give a bonus of 4 + Spellcasting (yes, 100% conversion of spellcasting to physics in one cast!), and haste gives a bonus of +10 x Spellcasting %.