Imagine this query in Postgres:
SELECT * FROM records WHERE ORDER BY timestamp_desc;
If there are 1,000 items in the database, whether you sort in Python or Postgres is irrelevant – the speed will be negligible.
If there are 1,000,000 items in the database, then database sorting is probably faster, as you won't have to do the operations in Python.
But consider this slightly different query where a limit is added, which would happen before the sort:
SELECT * FROM records WHERE ORDER BY timestamp_desc LIMIT 100;
You need to sort in the database, so re-sorting in python is largely irrelevant – but if you do that, it will likely be negligible.