[rdonj]

I just want to point out one more thing - you can't really think of effectiveness of cast spells in terms of how much gold the mage casting it cost you to purchase. I.E. 112.5 gold per thunderstrike is a bogus figure. The real cost of using mages in battle is that they aren't spending their time researching better spells instead.

But even if you want to look at it in terms of gold cost of a mage casting spells, one mage makes back 100 gold for each random AI soldier you kill, approximately, generally more because the AI does recruit expensive soldiers too. But anyway, each battle a mage is in he kills more soldiers. It is incredibly easy for any random mage to get more than 50 kills over the course of their lifetime serving on the battlefield, especially against the AI. Once you've made up for their initial purchase cost of the mage after probably as few as 3 battles, then you are just paying for his upkeep cost, which is easily overcome by kills made. You should only lose mages fairly infrequently, and primarily due to stray arrows even then. This is important because the lifetime effectiveness of said mage should quickly outstrip the gold effectiveness of purchased troops who will die a lot more often and force you to keep spending the premium up front cost of the purchase.