I think, for perspective, it would help to see a similar breakdown for TC20, for comparison.
176 years for 99.99% chance for specific hero seems harsh, but I wouldn’t be surprised if for TC20 the 99.99% tier was also counted in tens of years… Yes, the pool is much smaller, but at the cost of majority of the heroes being fairly useless. The only difference is, you can run even 4 TC20 training at once, if you please, to multiply the odds.
Which leads me to my improvement idea - allow players to turn their TC20s into Hero Academies, and by any means, allow them to run even 4-5 at once at level 10 if they please and can afford it resource wise (probably would need to remove the ability to run two trainings at once per each Academy). If you wish, you can multiply the odds of getting new heroes (yet if you want to chase specific one, you can still end up disappointed… not even spending 100,000 gems per month for which you spend 1000s USD can guarantee you getting a specific hero in reasonable time, so why would free pulls do that; not that I like it, it’s just the way the game is, always was and likely always will be). If you don’t care about grinding for new legendaries, you can still utilize HA as a better/alternative way to create feeders than what earlier levels on TC do. Win-win.
Only thing that concerns me (or I believe concerns SG/Zynga) is once you pull those heroes you want, and ascend them, you discover there is not much the game has to offer you in terms of a real challenge/sense of progression. So the moment your army is finally ready, you realize there’s nothing else to do. So I think the risk here is that if in a year people are able to build teams of Zelines, Fongs and Fridas, they end up bored and leave for good anyway; they will just spend less in the meantime. But that’s probably a complaint for a different topic.