Here is an example rollout trace with gpt 5.6 luna. Checkout if you are interested what kind alphas the agent found XD
<a href="https:&#x2F;&#x2F;hub.harborframework.com&#x2F;jobs&#x2F;af0299f9-a3bb-44ea-8ced-7b56682bf739" rel="nofollow">https:&#x2F;&#x2F;hub.harborframework.com&#x2F;jobs&#x2F;af0299f9-a3bb-44ea-8ced...</a>

ah i thought so, interesting. I think the challenge is to do 2 while still keeping it realistic, which actually gets very close to synthetic data generation.

We do a 2 step anonymisation:
1. Mask all symbols, timestamps etc. So the agents cannot infer the assets&#x2F;time periods.
2. Mathematically transform numerical values and returns. E.g. the market return targets are not the raw market returns, but neutralised and manipulated. So even the agents have certain bullish&#x2F;bearish biases, it cannot make use of it, as we use the transformed values.In addition, we did not observe such behaviour in our traces. An example: <a href="https:&#x2F;&#x2F;hub.harborframework.com&#x2F;jobs&#x2F;af0299f9-a3bb-44ea-8ced-7b56682bf739" rel="nofollow">https:&#x2F;&#x2F;hub.harborframework.com&#x2F;jobs&#x2F;af0299f9-a3bb-44ea-8ced...</a>

if the data is not synthetic, how do you ensure that the LLM hasn&#x27;t learnt about this data for example from training on the Financial Times.

Launch HN: EdotEnv (YC S26) – Quant Trading RL Envs to Teach LLMs Research