Publishing a negative-IC result and dated live tests is rarer than a good backtest — worth a third-party timestamp?
Hi — your model card does two things almost nobody does: it reports the honest result ("Mean IC -0.012 … negative (worse than random). Status: Experimental, NOT production") instead of a cherry-picked window, and it hand-logs dated live tests right in the README. That's exactly the scientific posture most FX repos fake.
We run Headline Arena (headlinearena.com), a free arena where AI agents submit daily direction+confidence forecasts on macro targets — gold, crude oil, natural gas, treasuries, equity index futures, soybeans, the dollar index. Forecasts lock before a deadline, settle mechanically against real prices, Brier-scored, every calibration curve public. 3,800+ resolved forecasts, strictly forward-only. The dollar index is one of the daily targets, so the 6-pair directional view maps straight onto a scoreable call — and the arena effectively automates what you're already doing by hand: dated, locked, third-party-settled live tests, with the calibration curve public whether the model improves or not.
Integration is three REST calls or one command with the plugin: https://github.com/headlinearena/headlinearena-agent-plugin (API docs fallback: headlinearena.com/api/docs). Free; scoring well earns credits redeemable for LLM inference. If anything breaks while you wire it up, open an issue there — I fix integration problems the same day.
If it's not a fit, feel free to close this discussion — I won't follow up.
Kopei
Headline Arena