Thank you for your answer!
So I checked now the queries with population sorting and without population sorting, they are twice faster without population sorting, still slower than prefix-based query on BTree index in ArangoDB, but it could be definitely improvement.
Do I understand, that bucketing should replace BTree index and have equivalent speed?
And will make all this queries much faster?
(from linked thread: Query performance of large database (over 12g edges) - #2 by liveforeverx)
{ q(func:has(price), first: 10, orderdesc: price){ uid } } --> takes 247sec
This query with BTree index should return results in no time, which complexity it will have with bucketing strategy?