orko | script multiple crypto exchanges from a single user | Cryptocurrency library

It seems the only shared state you have in the whole thing is leaderBoardUserEntryMap. You synchronize on it when updating in sortLeaderBoard. In getLeaderBoard for some reason you don't yet synchronize on it when checking if (!leaderBoardUserEntryMap.containsKey(uid)). That's minor, but you should do it as well. Later, you synchronize on it when constructing the leader board.

From this point of view your synchronization seems adequate.

What I find a bit problematic is that your Entry is mutable and stores position. This makes your update operation more problematic. You have to re-sort and re-set positions on every update. And you're locking all other update or even read operations. I'd avoid mutable objects in multithreaded code.

I'd use a SortedSet instead and let the implementation handle sorting. To find out the position of the element you'd just do set.headSet(element).size() + 1. So no need to store position at all.

I initially wanted to suggest using concurrent collection implementations like ConcurrentHashSet which would allow "full concurrency of retrievals and adjustable expected concurrency for updates". Basically, retrievals could be non-blocking, only updates were.

However, this won't help much as your "retrieval" logic (creating the leader board around target entry) is not so simple and involves several reads. So I think it's better not use concurrent collections but instead actually synchronize on the collection and make updates and retrievals as compact as possible. If you give up on the idea of having position in Entry then update is a trivial add. Then you'll only need to read entries around the entry as fast as possible (within the synchronized block). This should be quite fast with tree sets.