alt Hacker NewsI don’t understand this constant fascination with having language models trade stocks. Language models are very useful tools but not aligned at all with generating alpha.
Replies
I use the Interactive Brokers MCP pretty heavily. I don't do any cool automatic fun "trading", but instead I use it to have "pseudo-QQQ".
I didn't like the relatively high fees for QQQ, and I realized that Invesco releases the weights for QQQ for free. I also think Tesla is too overvalued, and I want to avoid the SpaceX IPO. With the Interactive Brokers MCP, I just feed it the CSV of QQQ's weights, tell it to remove and redistribute Tesla, and then I tell it to buy "$1000 of pseudo-QQQ", in the form of raw stocks.
Doing this, I still basically get the same exposure as QQQ, without any fees.
They’re great at generating alpha, just not for these users.
An LLM may be bad at trading stocks, but an LLM may be good at analyzing the wider context, like the news feed, to inform automated trading driven by a more sophisticated model, called by the LLM as a tool.
I don't think that this contraption should necessarily perform tolerably, but the use of an LLM is not necessarily a wrong move.
As much as I hate the idea of enabling the desperate masses to gamble like this, LLMs are very aligned tools for sentiment analysis, which can be the foundation of a trading strategy. I think it's extremely irresponsible to use them for execution, though.
The usual question: what's "aligned with generating alpha" that a human stock trader can do, but an ai can't?
AI agents for trading, as well as 24/7 trading are no different than offering sports gambling and prediction markets to the masses; it is a vacuum for the fiat of the unsophisticated. The goal is more trading volume to generate more fees, similar story with private equity wanting access to 401ks to unload PE at peak valuations to bag holders.
Alpha is ultimately the result of analysis, of better analysis than others.
LLM's can actually be exceptionally good at research and pattern recognition, i.e. analysis. And while they aren't great at running numbers themselves, they can do exceptional work passing off Python scripts to an interpreter to generate the numerical results they need.
I'm quite sure the Robinhood AI is going to be trash, i.e. just a gimmick.
But, it's not crazy to think that with the right harness, there are big opportunities for identifying profitable strategies. Especially relying on unparalleled and essentially unlimited research capacity based on public information. More analysis than any single firm could ever hire.
And even for Robinhood users, it's entirely plausible that AI-traded stocks will perform much better than the trades a majority of users would make, since most investors are really unsophisticated.