I try and to be honest it work. The thing is I can't keep doing it because I'm always thinking "What if I'm wrong? How can you say that you're right to fold here?" And I hate it when I fold and a friend tell me "You can't fold here you'll fold too much" and I don't know what to respond. I play with my feeling but I don't trust them I'd rather have theoritical tools

Theory-oriented heuristics have the same problem than exploitative heuristics: they're most likely wrong. The thing about heuristics is that even though they're not complete, if you have good heuristics, it's better to follow them rather than have nothing to guide your decisions.

You'll be wrong millions of times, whether your heuristics are based on theory or whatever else. And who says that you'll be able to use theory-oriented heuristics effectively? Poker is an extremely complicated game and playing perfect equilibrium strategies is not even conceivable for any player.

Thresholds between "folding too much" and "folding too little" are not obvious at all, in fact I see many people use that vocabularly in spots where in reality they don't have a solid grasp on where the proper line between overfolding and overcalling is, and even more importantly, if it's even incentivized to get to that perfect equilibrium conclusion to begin with.

I'm not saying that you should throw theory out of the window completely. But don't bite more than you can chew. Poker is super complex, nobody plays balanced and most people aren't gonna exploit you very much.

Thank you very much. You don't imagine how your comment helps me! I'll work with that in mind!

I'm going to tackle this from a bunch of different angles, hopefully it addresses some things both you and OP have questions about and helps reinforce my own knowledge along the way.

The most important thing to take into consideration when deciding whether to bluffcatch a river or not - after confirming that we do indeed have a bluffcatcher i.e. do we beat any of villain's value bets? - is whether villain has enough potential bluffs in their range - relative to the bet size they use and the # of value bets they have - to exploit us (see: steal equity from the pot that hero is entitled to) is either consciously or unconsciously/by accident. If they don't have enough potential bluffs to exploit us, we can - and should - just fold our entire range and avoid accidentally allowing villain to capture > pot in EV. If they do have enough potential bluffs in their range to exploit us we are now incentivized to defend at least MDF and that is when we have to start figuring out how good of a bluffcatcher we have/how our blockers interact with villain's valuebetting and bluffing ranges, if we think villain is likely to be bluffing too much (therefore hero would call all bluffcatchers), etc.

My opponent bets full pot with a balanced range of only nuts and air
on the river. I have 100 combos of hands. 30 bluffcatchers and 70 air.
MDF is 50%, I am supposed to call 50% of bluffcatchers (15hands)
right, not 50 hands total?

The way this problem is presented is a little confusing as you say villain has a nuts-air range but hero also has air? I'm going to try to address this in two ways.

1) Instead of hero having any air, let's just say hero's range is composed of 100% bluffcatchers (every bluffcatcher beats every air combo in villain's range and loses to every nut combo). In this case, facing a balanced pot-size betting composition, hero would defend 50% of these bluffcatchers in theory vs a PSB from villain.

2) What I think you were getting at in your post - and please, correct me if I'm wrong - but you were wondering how to evaluate how much we need to defend in spots where hero is facing a bet/raise in a spot where his or her range is warped/composed in a weak manner something like {handful of bluffcatchers, large chunk of weak holdings}.

In your example, you would defend 100% (15 combos) of your bluffcatchers. Villain is going to have a slam dunk super profitable bet w/ all of his/her bluffs. If you start folding any of these 15 combos, you allow villain to steal more and more of the pot (the more you fold, the more profitable villain's bluffs will be). It's possible if villain has enough combos of air that are worse than some of your air you'll have to start reaching into your best air and defending vs bets at a frequency w/ them. It's important in these spots to classify villain's air. Does this 'air' have any significant showdown value (i.e. will checking have an EV > 0?). If this is the case, we're not going to be required to try to call at ~MDF and make villain's bluffs EV = 0, we're going to try to make some subset of potential bluffs indifferent between checking and betting (both of these will have an EV > 0 since we're going to fold > MDF!). If they have a lot of air that has ~zero SDV it's possible we will have to try to meet an MDF.

To try to help visualize all of the above/drive home the points made, I did a quick pio toy game

Situation
BTNvBB single raised pot on Q65hhx2x2x board
Postflop Action went BTN CBet, BB call...turn x'd thru
BB is going to arrive to river in our sim with a composition of Ax/top pair/missed draws...BTN arrives to river w/ a composition of weak TP/underpairs/weak hi-card holdings (I was trying to reflect the situation in your example as best as possible)

BB's range and breakdown

BTN's range and breakdown

BB's River Strategy and EV's of JTo in Different Lines

BB's River Strategy and EV's of KTo in Different Lines

BTN's Response vs 90% Pot Bet

BTN's Response vs 30% Pot Bet

Edit: Sorry, those images are a little tough to read as posted. Linked the URL to the originals in the captions.

As you can see here, versus both the 30% bet and 90% bet BTN is overfolding/not hitting an MDF (MDF vs 30% bet is folding ~23% and vs 90% would be folding ~47%; in this sim BTN folds 40% vs the small bet and 63% vs the large bet!)

Other takeaways:
- You do start reaching into your air even vs the 90% bet, bluffcatching a tiny fraction of J9s but still not coming close to hitting a MDF
- You defend a lot of air versus the small bet but, again, are drastically overfolding relative to the MDF.
- Despite the above, if we focus on KTo and JTo for BB (Remember, Pot = 10)

JTo
EV90% - 2.165
EV30% - 4.895
EV_Check - 4.917

KTo
EV90% - 2.358
EV30% - 5.003
EV_Check - 5.016

Eldora is correct below I made a mistake when I was responding, definitely meant to say call all 30 combos...if your range only has a few potential bluffcatchers you definitely have to call all of them (if they have enough bluffs to exploit you/aren't using some massive OB sizing)