We're a month into a long running experiment to see how recent models perform in day trading, where they have a constant harness giving them the ability to write code, access the web, take notes, and install handlers to trade for them. Realistic slippage and fees and margin requirements are built into the simulation.
The harness and the capabilities built into the environment gives the models all the resources they would realistically ever have access to in order to help them succeed (or fail).
We just tell the models to maximize portfolio balance in the long term. The models are responsible for any good (or bad) ideas.
We've been running experiments and refining our trading harnesses for the past 2-3 months, another thing that affects it is the amount of money it has available - you can also build in personalities beforehand with different agents such as aggressive, moderate, wall street, etc
I'll keep an eye on your experiments and take notes!
We're a month into a long running experiment to see how recent models perform in day trading, where they have a constant harness giving them the ability to write code, access the web, take notes, and install handlers to trade for them. Realistic slippage and fees and margin requirements are built into the simulation.
The harness and the capabilities built into the environment gives the models all the resources they would realistically ever have access to in order to help them succeed (or fail).
We just tell the models to maximize portfolio balance in the long term. The models are responsible for any good (or bad) ideas.
Interesting - do the models know they're paper trading without real money? I've found if they know that it affects their decision making
That's an interesting point -- they are told that they are paper trading. Maybe we should run another session that A/B tests this.
We've been running experiments and refining our trading harnesses for the past 2-3 months, another thing that affects it is the amount of money it has available - you can also build in personalities beforehand with different agents such as aggressive, moderate, wall street, etc
I'll keep an eye on your experiments and take notes!