hard to weight specific patterns relative to each other properly because in that regard reading is super subjective, some people are good at stars but bad at squares, good at squares bad at stars, good at spaced streams bad at jumps etc
AR is an even bigger wild card, we know the meta is like ar9-9.6, that doesn't mean there aren't players who specialize in lower or higher, but that doesn't make them better.
and you can't really tell how good an ez player is vs a nomod player because of how different they are.
One thing you could try is to normalize everything on a map minus the AR and compare their performance on it, and even then it's not really fair because all the time the ez player spent improving in high density reading was spent by the nomod player primarily on acc or hitting smaller circles so it's a tradeoff either way. Just two very different playstyles... You can meet in the middle in AR/CS maybe, like if an ez player is used to cs2 ar4 and a nomod player is used to cs4 ar9, make the map ar6-6.5 and cs3-3.5? Only issue is they'll both be out of their comfort zone but hopefully equally so. Changing the OD will only disadvantage the ez player though so maybe keep those at their usual. Acc pp isn't intended to be rebalanced here anyway.
Just shooting ideas and how to help evaluate the mod. It's probably always going to be a penalty just due to the OD decreasing either way, unless some low ar reading pp bonus implemented, and even then, it should only be in the most difficult EZ maps that the bonus is enough to completely make up for the acc pp loss by the huge reduction in OD that ez brings.
Anything that results in a more accurate rank for ez players is welcome... Ekoro is laughably underrated.
Fix speed pp while you're at it