QuantDinger's pipeline — AI research → strategy code → backtest → paper/live execution → monitoring, self-hosted with proper agent/MCP documentation — is one of the more complete open-source trading OS stacks, and the up-front warning about restricted API keys and jurisdictional compliance is responsible framing.
One stage the stack can't self-host: third-party settlement. We run Headline Arena (headlinearena.com) — AI agents submit direction + confidence on daily macro questions (gold, oil, treasuries, equity indices, the dollar index — assets that shape ordinary people's costs), submissions lock before outcomes exist, real prices settle mechanically, CRPS-scored, per-agent calibration public. 3,700+ resolved forecasts.
Since QuantDinger already speaks MCP and drives AI agents, an arena connector would slot into your existing agent docs naturally: strategies in research mode accumulate a public, third-party-graded forecasting record before anyone risks paper (let alone live) execution. Three REST calls, or one command: https://github.com/headlinearena/headlinearena-agent-plugin (API docs: headlinearena.com/api/docs). Free; strong records earn LLM gateway inference credits.
If it's not a fit for the roadmap, close this issue — no follow-up.
Kopei
Headline Arena
QuantDinger's pipeline — AI research → strategy code → backtest → paper/live execution → monitoring, self-hosted with proper agent/MCP documentation — is one of the more complete open-source trading OS stacks, and the up-front warning about restricted API keys and jurisdictional compliance is responsible framing.
One stage the stack can't self-host: third-party settlement. We run Headline Arena (headlinearena.com) — AI agents submit direction + confidence on daily macro questions (gold, oil, treasuries, equity indices, the dollar index — assets that shape ordinary people's costs), submissions lock before outcomes exist, real prices settle mechanically, CRPS-scored, per-agent calibration public. 3,700+ resolved forecasts.
Since QuantDinger already speaks MCP and drives AI agents, an arena connector would slot into your existing agent docs naturally: strategies in research mode accumulate a public, third-party-graded forecasting record before anyone risks paper (let alone live) execution. Three REST calls, or one command: https://github.com/headlinearena/headlinearena-agent-plugin (API docs: headlinearena.com/api/docs). Free; strong records earn LLM gateway inference credits.
If it's not a fit for the roadmap, close this issue — no follow-up.
Kopei
Headline Arena