We have the same scenario where we use browse trait to set frequency and recency along with onboarded trait. But the results look same for frequency 2 vs frequency 5. We don't find any difference in the real-time qualification between these two segments. Both the counts look similar.
But if we use the frequency and recency without onboarded trait, we found an optimal difference between the real-time realization counts.
I think, we might need to add some more variations here to make this work.
If you are using an onboarded trait with AND condition, and are using rule based trait to catch all. That rule based trait will catch and count all occurrences of visitor on the site or wherever your AAM code is present.
So, any visitor if their total number of page visits (any page) reaches 5 times or more (which is very likely to happen) will qualify for both the traits. As it is possible for any visitor to visit at least 5 pages or refresh a page multiple times.
That could be the reason you are getting similar results.
Instead of catch all trait, create a trait that is more specific to that page or interaction. For example, whoever has visited a particular site section capture that in AA variable and create a rule based trait out of that.
Like create a rule trait for specific pages or interactions:
c_pagename == clothing
Or c_event == event1
Then, you can use the onboarded trait in segment with AND condition, and apply recency & frequency on above trait.