As far is a known the retry with backoff strategy is an solid one because the rate limit is determined by spl, which is necessary for them so the api traffic is not overloaded.
You can look if you can make the api call bigger so it includes more data at once (less calls). But i expect you need to do something for every brawl cycle.
These are the reason (for me to retrieve many transactions and balances) why there is sometimes a large wait to retrieve data from spl.
Not for your case but if you store the already retrieved data you do not have to retrieve it again in a later phase. Then you have to think on how to store it, this is where the application get more complicated imho.
RE: Updates to spltools. Classes for Brawl data