In case there is interest, I've got a fork with some custom improvements like timeout handling, date validation, stop-loss direction checks, and extra utility helpers.<p><a href="https://github.com/skanga/TradingAgents" rel="nofollow">https://github.com/skanga/TradingAgents</a><p>I'll also be merging the upstream commits ASAP
103K stars, so clearly it's popular. Has anyone here used it, and what are the outcomes like, and importantly, who is the target audience for this?<p>I can see the intention behind crawling social media and news feeds to determine some 'evidence', but am not sure if that's the best approach or even if an LLM is the best way to get an assessment, or whether having so many input sources is a good idea.
I do not understand the value of multi agent approach? Isn't a single agent with a good harness better than any multi agent env?
It helps you spend more token, is more expensive, and thus is obviously more AI. Also novelty and more complexity means less scrutiny of the approach.<p>These are necessary and perfectly sufficient for an investment firm thesis I believe.
I think multi-agent (eg _different_ underlying LLMs) everything is really the future. Code produced via multi-agent workflows and reviews seems noticeably better. I've been experimenting with a multi-agent message board recently: <a href="https://github.com/pjlsergeant/dogpark" rel="nofollow">https://github.com/pjlsergeant/dogpark</a>
nightmare horseshit, don't waste your tokens