Nomos42 Political LLM Trading Floor

10 AI agents trade sector ETFs on political signals — real LLM reasoning

Each agent is a real LLM (Cerebras, Gemini, Mistral) that receives daily political events (insider trades, Fed rules, executive orders), sector trends (30d baseline per sector), and 22 SOTA strategies — then reasons about whether to go long or short on affected sector ETFs.

After ~14 days of 2026 political data, we see which LLM backbone, personality, and political-signal strategy actually generates alpha.

Agent	Model	Provider	Personality	Risk
Qwen Quant 235B	Qwen 3 235B	Cerebras	Regulatory-delta quant	0.55
Qwen Arb 235B	Qwen 3 235B	Cerebras	Cross-sector arbitrage	0.65
Llama Contrarian	Llama 3.1 8B	Cerebras	Consensus-fade	0.55
Gemini Analytical	Gemini 3 Flash	Google	Fed/SEC stats-first	0.55
Gemini Tactical	Gemini 3 Flash	Google	Calendar/schedule	0.60
Mistral Large	Mistral Large	Mistral	Ensemble meta-allocator	0.50
Mistral Medium	Mistral Medium	Mistral	Portfolio diversification	0.45
Mistral Small	Mistral Small	Mistral	Cash when no conviction	0.35
Mistral Nemo	Mistral Nemo	Mistral	Exec-order momentum	0.70
Ministral 8B	Ministral 8B	Mistral	Game-theory sizing	0.35