Your argumentation has holes. First, the stats are by far not reliable since they just judge a some games by civ, but not by strategy and such. And even if we trust them and assume you face an equal player every time, then why is 6/100 games you played byzantines (= 3500 games randomized) a bad thing ? And then is the question still, did you choose the right strategy for that matchup or are you still unfimiliar with all nuances the civ has to offer.
And you skipped in your argumentation the most important percentage: Arabia 1650+, their probably worst map, played by people familiar with the game and not people that just played three civs to grind up their elo. 49,7%
In what world is that unbalanced ? And if you look at their winrate over gamelength then they are by far only crap in early game and super good in late game
People just learn 22 pop scout rush Buildorders and try it with every civ, thats why you cannot use low elo stats for balancing, and hence not averaged values because low elo ofc dominates the played games. But anyhow with 200 games rated, the stats are somewhat open for discussion and it must also be considered if the civ has real weaknesses and in fact, byz dont have those, they are on any map at any stage of the game better than C-Tier. Hera is just a biased player sometimes that hates slow defensive civs and is not really the go-to guy for judging the game. Its not like pros are better game designers.
I know, you might have fallen in love with your idea, but I dont see any point in adding it, other than it changes anything for the sake of changing and you being happy having had that idea. And yeah, with that I like to end my point here, maybe you still like your idea, then go on, I know the fun of game design questions as well, and I dont want to ruin it.