hyperc-ai/p34-technical-report — reverse-engineered prompt

I want a small research demo that shows this P34 trading agent idea in a way I can run and understand.

Please make a reproducible notebook centered on a synthetic partially observed market, where a normal profit prediction model looks good in training but fails when it has to act on its own. Then show the P34 approach making better trade decisions by skipping the false positives that hurt the baseline. I want the slower market waves example, clear charts of cumulative profit, and some kind of replay animation so the difference is easy to see.

Keep it framed as research and engineering only. Add very obvious disclaimers that this is not financial advice and that the results are synthetic, not live market proof. It would also be great to include a simple written discussion, like a technical report plus a shorter customer friendly brief, explaining the problem, the benchmark setup, and the results in plain English.

Make it easy to run locally, and look up current docs online if you need to.