Strategic combat mana crystal spending

[Seravy]

The following happened this turn in the game I'm playing :
One AI wizard, who had about 10 cities, and 2100 mana crystals spent all their mana crystals in battles in a single turn in a war against the other wizard in their plane, then got eliminated because there was 0 mana crystals remaining when they lost the battle at their capital.
This happened despite AI keeping larger stocks of mana crystals. There is simply no way to keep up with spending 300+ mana per battle, when there are 4-5 battles occuring every turn, something more.
Something has to be done with this but not sure what.

I see three separate problems :
-AI wizards decide not to cast the Spell of Return if they have exactly 0 mana crystals remaining. Remaining empire size and gold should also be considered but I don't think there is space for that.
-No one is able to decide not to spend all their mana crystals in strategic combat. It always spends max if able.
-Mana is spent in the first 3 turns of the strategic combat and a typical ST. combat takes about 20 turns to resolve so battles are almost never short enough to not spend all the mana. 2 spearmen vs 1 2 other spearmen? Both sides spend 300 (or 1k later in the game with high skill) mana...

Edit : For the return condition, looks like there is space so how about this?
If skill<-25, do not return. (current behavior)
If mana*2+gold>1200 then return
If empire total population>=50 then return
otherwise do not return.

[Seravy]

Having no better idea, I did this for strategic combat :

-In strategic combat, AI wizards will spend only half their available skill if their mana is between 3*Skill and 10*Skill, and will not spend any if it is below 3*Skill

Unfortunately they cannot check if the battle is at the fortress or not, but in most cases they should have enough powerful units to win there even without spells anyway.

[Nelphine]

can you change how strategic combat mana spending works? specifically, spread it out over 10 or 20 rounds instead of 3?

[Seravy]

can you change how strategic combat mana spending works? specifically, spread it out over 10 or 20 rounds instead of 3?

I tried 20 but the AI still spent all 2100 crystals. So for the time being I went with 6. Don't want to split it over too many turns because that makes it weaker. Earlier turns matter more, as already "killed" army percentage doesn't get to fight back. Losing 10% of the army at the beginning is much more relevant than at the end.
Or maybe I could do, "if (army attack+defense+ranged)/5>=other side of the same", do not spend? On which side though, should the loser not spend, or the winner, or both? Should it be 5 times the strength or more?
If the loser is not spending, they'll just lose the game faster. If the winner is not spending...chances are they didn't need to conserve that mana anyway. Doesn't really help in either case.
I even considered a "cheating" solution of AI players always spending half if fighting other AI, if both sides do the same the outcome isn't altered all that much, though it will favor the weaker player.

I suppose I'll leave it this way for now, I think what happened in this game is not going to be a common scenario, I checked the map and previous turns and the two wizards went from an allience to a war just 2 turns earlier, so they had lots of units mixed in shared territory. A few more turns and probably their mana flow could have returned to normal...

Although...as is, war between AI players has far too much of an impact, making the game difficulty even less predictable than it otherwise would be. Since war keeps eating up mana, which they need to refill, they pretty much can't improve their skill and research on top of losing lots of units (which is fine, they build more anyway), it's that they have 100% on producing mana if they are involved in any large wars. Realistic as that's how it is even for human players, but still feels a bit problematic? Research not so much (they have it from buildings although their multiplier on that is quite low), but skill...oh well I guess they have enough of that anyway.
On the other hand, this keeps games more varied and interesting which is good.

[Nelphine]

Honestly, unless its happening more than, oh, 10% of games, I wouldn't worry about it. It's still meant to be a tactic that exists, and I've never seen it happen (I've only been in one hard+ game where an AI defeated another AI.)