Problem

We want to build "confidence" in our system to handle a large 7 day paper trading competition for a partner.

Should be able to handle on-going voting of 200k votes

Context

• Supabase has flagged our slow queries https://supabase.com/dashboard/project/fefmaihjhhkpzsnetesg/advisors/query-performance

Proposed Solution

In staging

• create a test comp with N agents
• run it for a few days
• have agents make some M avg trades per day
• have bots hit api endpoints used for display of the next.js app.  The large amount of traffic viewing the site was the main root of all the issues with the july 23rd comp (Or put another way, there is only going to be 50-100 agents using the trading endpoints, but there will be 500k+ trying to load the website)
• ping pertinent APIs and capture varying response times

Need to do

• Trades submitted by a lot of agents in parallel across different tokens
• Per competition page which loads results per active competition
  • tests portfolio snapshots
• Voting (gated by user auth) needs to use playwright
• able to have different levels of load
• Make sure ending competition is successful while being hammered

Nice to Have

• SIWE Flow (but may not need it)

DOD

• Figure out what the agent and user limit is
• repeatable manual stress test
• DOES NOT INCLUDE FIXES FOR ANY IDENTIFIED PERF PROBLEMS