In a format where the vast majority of regulars dont have fixed preflop sizes what makes more sense when we want to study the reactions against 3bets

Assuming we are facing 3bets 6bb 7.5bb 9bb

1) Run a sim with multiple 3bet sizes of 6bb 7.5bb 9bb and react to the size that is the most close. We are running into danger that one of the sizes will never or rarely be used and the response will be completely off since the range will be 0.01-0.5% frequency
2) Run a sim with the average gto size, assuming for now that is 7.5bb, then take the 7.5bb range that we found and lock the same range for the other two sizes in two different sims
3) Run three different sims with 6bb 7.5bb 9bb each

Whatever sim models best what your opponents are actually doing with their range.

Ultimately it's impossible to learn hundreds of responses by heart, so I'd memorize the one vs main sizing and then look at a few others to get a sense of how much It shifts

i dont understand why you bother replying with something thats completely irrelevant to the question

1) everyone knows is impossible hence i ask the best methodoly
2) then you proceed by you will look a few other without even answering the best methodoly in your opinion

Your answer is competely irrelevant to this thread

Who hurt you Buddy

Also reread first sentence

which is impossible to know with the exact frequencies they are using buddy

No one talked about frequencies?

Are your opponents RNGing multiple sizings -> multiple sizing sim response
Are your opponents using one sizing with their entire range, and changing their sizing based on some variable -> single sizing sim response
Are your opponents using different sizings based on their hand disregarding balance? -> Use your brain and research how they're messing up, exploit it

This is actually an "exploitative vs GTO" question in disguise.

Aner0 suggests using whatever sim best models their actual strategy, which is designed to maximize against someone who only uses that size.

The alternative is to give villain all 3 sizes simultaneously and measure how the solver responds to each; this is less exploitable and is closer to the true GTO answer.

----

To make this easier to understand, I suggest developing pricing elasticity charts. I go over that process here: