Not familiar with Oracle's internal components, but here is my understanding of how this works with Postgres:
Indexes are clustered for all purposes and purposes. Therefore, if you are ordered that new lines are added at the beginning and at the end of the line (for example, created_at, updated_at, billed_at, etc.), your new lines will be added (or almost like that), and not added (leading to the disk page splits). It's faster.
Your query planner will happily read the index in reverse order. Therefore, if this is a single column index, or it works, use the most natural one when it comes to how new rows are inserted into your use case.
If an index, ordered in reverse order, can become interesting, this is when you have an index with multiple columns. Say (id, created_at desc)in the audit log table. This is actually a bad example, but here's the point: if you order id, created_at desc, the index will be used as is.