Hey there, sorry about the bad luck recently. I always try to take bad luck moments to double down and work harder so that by the time I'm running better, I'm that much stronger of a player.
As for what type of under EV you can experience, it can get rough at times. You could see 50-100 buyins below EV.
The good news, though, is that one game of hypers is only 3 minutes or so on average, meaning you can play tons of games in a short amount of time and your downswing can go away quickly.
I was once down around 10k in $200-350 buyin super turbos back on Full Tilt. I ended up winning around 18 in a row during a session vs one single player and making back so much of my money in just a simple 2 hour session.
So things can change really quickly in this game, it's best to just focus on your own decision making and try to make the best play every single step of the way and not let emotions get in the way of finding the correct answer to each spot.
About EV, it's not perfect (there's nothing that says we're unlucky bc we ran into the top of an opponent's range, for example if we have KK and our opponent has AA, our expectation is pretty low, meaning we expect to lose, but if we reversed the cards we were dealt our opponent would obviously get it in in a hyper with KK with a smile on his face...). You can extend this theory (card/hand distribution) to the flop, what sort of flops are we getting compared to our opponent? Unfortunately, we can't really ever know if our opponent is getting worse flops or turns or rivers than us, bc we don't know his hand and exact strategy every step of the way (it's too hard to even accurately tell if we are running bad in many situations too).
All that said, EV is better than results, it removes some (small) element of luck from the results on average I believe, making it better to pay attention to than your results.
In reality, you really just want to focus on always improving, especially at the smaller stakes games.
Good luck and let me know if you have any other questions, thoughts or updates.