Espers teaching Stat gains

Cool! I was hoping somebody would respond. Thanks! I think what might be happening is that I know just barely enough about romhacking this game to misunderstand everything about romhacking this game. LOL.

Stat gains are limited to one at a time and although they can stack, the code is designed to increase one attribute per character upon levelling up. By contrast the varying nested subroutines concerning learning a spell are designed to work with the spell index to first skip a spell if it is learned, then mark a spell as just learned and to mark the spell as unhidden/accessible. There is no mechanism for learning a spell over and over again, so the subroutine you would eventually create would have to be completely new.

Oh, I didn't have any interest in increasing an attribute multiple times or learning the attribute over and over again. It would work exactly the same as learning spells. Once you've gotten Shiva's "Speed+2...x4" or whatever the skill is to 100%, that's it. It's at 100%. You'd get Speed+2 permanently, but it wouldn't start over.
Regardless, you're still totally right that I'd need to build a new subroutine. The tantalizing part is that so much of the existing code is already doing what I want! The only part I'm looking to change is the result. If I better understood the moment the code says: "Alright. They got to 100%. Go flick the on/off switch in the RAM to make that spell available", I could potentially intercept it and push it to a new subroutine that goes to the stat gain formula instead. At the moment, I don't understand how to do that. I mean, aren't the percentage points just a value stored in the RAM that's counting to 100? How "tied-in" is that number to the spell index?

To put it in FF terms, my coding level is also "too weak to be worthwhile". I suppose I'm not doing a great job of explaining my confusion.

Creating your own stat gain index is an interesting work around but you would have to reconfigure every single loop concerning learning magic from Espers to account for that index. 

This was my idea for a second option if the above "intercept the moment of spell learning" was functionally impossible or just straight-up dumb. We know the Espers have spell index values in the Esper data at $D86E00. I thought maybe creating a new index for stat gains (a ton of additional work) might be a good Plan B.