Scored some
@Bitget_AI AI Hackathon projects today.
Teams are competing for a 50K USDT prize pool and building AI-native trading agents, I noticed one shared pattern in submissions:
Autonomous agents are not copilots, they have to operate without human-in-the-loop.
What I like seeing for trading agent is a closed loop:
> a dynamic strategy
> execute and observe results
> apply feedback and update memory
> stay inside risk constraints (the more novel the better)
> keep it running