News for hackers. Stuff that matters.

A comment on We gave 5 LLMs $100K to trade stocks for 8 months. Parent comment

Re:We gave 5 LLMs $100K to trade stocks for 8 months by plufz on Thursday December 04, 2025 @06:20PM (#46154626)
I know very little about how the environment where they run these models look, but surely they have access to different tools like vector embeddings with more current data on various topics?
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by endtime on Thursday December 04, 2025 @06:52PM (#46154982)
If they could "see" the future and exploit that they'd probably have much higher returns.
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by alchemist1e9 on Thursday December 04, 2025 @07:16PM (#46155191)
56% over 8 months with the constraints provided are pretty good results for Grok.
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by plufz on Friday December 05, 2025 @03:06AM (#46157897)
I would say that if these models independently could create such high returns all these companies would shut down the external access to the models and just have their own money making machine. :)
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by disconcision on Thursday December 04, 2025 @06:44PM (#46154903)
you can (via the api, or to a lesser degree through the setting in the web client) determine what tools if any a model can use
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by disconcision on Thursday December 04, 2025 @06:55PM (#46155017)
with the exception that it doesn't seem possible to fully disable this for grok 4
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by alchemist1e9 on Thursday December 04, 2025 @07:16PM (#46155193)
which is curiously the best model …
[ Parent | on HN ]
Re:We gave 5 LLMs $100K to trade stocks for 8 months by plufz on Friday December 05, 2025 @03:08AM (#46157910)
But isn’t that more which MCP:s you can configure it to use? Do we have any idea which secret sauce stuff they have? Surely it’s not just a raw model that they are executing?
[ Parent | on HN ]