Yes, this is one of the methods, that my coach suggested.
To have 2 different chipEVs for 300 and 500 games and calculate breakeven points for them separately, and treat them separately.
I still have questions about what this "break-even" point means. Following this aproach as 500 chips games are 5x+ multipliers, your "break-even" point should be much lower (even with negative chipEV) as you get a lot of money when you win. And in 2x-3x multipliers you should win more to break even, following this logic.
There is a problem with PT4 labeling games strangely, so to separate 300 from 500 chips games - I do it manually after each session, marking all games as Ultra-turbo and then using filters to find the one's that we have dealt 24+bb effective stacks 3-handed and marking them Turbo.
In reality, I have next chipEVs for 100s
300c games: 16
500c games: 83
Spoiler


So the logical conclusion here is that I'm losing in fast games and compensating it in slow games, that makes sense as it gives more opportunity to realize your equity.
Before this suggestion, I used "weighted average". I got probabilities of 300c and 500c games, calculated what would be an average starting stack in games (318 chips for 100s limit) and used it in the formula that @dev209 shared.
I asked this question earlier in my blog marking it as "a million-dollar question - how do you calculate roiEV in 888poker Blasts?" but didn't get an answer)). Maybe we can come up with a good method now.