i ran a simulation on 28 paired boards on GTO + and the boards either fell into 3 categories(roughly) bet sizing used 1/4.

1. bet entire range
2. 50/50
3. check 60 bet 40

when you bet entire range , you either have either have a significant nuts advantage or both layers have the same amount of sets since you have so many overpairs.

when its 50/50, the BB has the nuts advantage (usually middling paired boards) and there is alot of draws and straight draws. so by betting here we deny equity to the BB's floats( draws), but we are betting less since they have the nuts advantage

when its 60/40. the board is still majority middling paired however there is less draws, so we want to be betting less since when we get called were likely beat, however we still want to bet since the BB will still over fold in this spot.

so developed some heuristics which are that as follow,

when we have the nuts advantage bet range

when we don't have the nuts advantage, on middling boards with draws, bet your nuttish hands and everything else 50% of the time

when we don't have the nuts advantage, on middling paired board with no draws, bet our nuttish hands and everything else 40% time

i node locked this strategy across the 28 paired boards and the EV loss i saw was 1.5%.

does anyone have better strategy to use here? was my analysis right?

You won't get much feedback on this without sharing more data.

What ranges were you using? stacks/pot? Bet sizes/betting tree? Was the solver allowed to exploit your locked ranges? Have you compared the EV loss to something stupid and lazy, like range-checking all flops?

In general, I've found that a small bet with a wide range is most effective on paired boards.