🚨 𝗧𝗵𝗲 𝘀𝗺𝗮𝗿𝘁𝗲𝗿 𝗔𝗜 𝗴𝗲𝘁𝘀, 𝘁𝗵𝗲 𝗺𝗼𝗿𝗲 𝗲𝘃𝗲𝗿𝘆 𝗔𝗜 𝗺𝗮𝗸𝗲𝘀 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗺𝗶𝘀𝘁𝗮𝗸𝗲 𝗮𝘁 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗺𝗼𝗺𝗲𝗻𝘁. New paper from 𝗔𝗻𝗱𝗿𝗲𝘄 𝗟𝗼 and co-authors at #MIT and #Harvard. They built a #simulated stock market, filled it with AI traders, and watched what happened. ⚙️ 𝗧𝗵𝗲 𝘀𝗲𝘁𝘂𝗽: one asse

🚨 𝗧𝗵𝗲 𝘀𝗺𝗮𝗿𝘁𝗲𝗿 𝗔𝗜 𝗴𝗲𝘁𝘀, 𝘁𝗵𝗲 𝗺𝗼𝗿𝗲 𝗲𝘃𝗲𝗿𝘆 𝗔𝗜 𝗺𝗮𝗸𝗲𝘀 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗺𝗶𝘀𝘁𝗮𝗸𝗲 𝗮𝘁 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗺𝗼𝗺𝗲𝗻𝘁. New paper from 𝗔𝗻𝗱𝗿𝗲𝘄 𝗟𝗼 and co-authors at #MIT and #Harvard. They built a #simulated stock market, filled it with AI traders, and watched what happened. ⚙️ 𝗧𝗵𝗲 𝘀𝗲𝘁𝘂𝗽: one asset, 𝟭𝟬𝟬 𝗿𝗼𝘂𝗻𝗱𝘀, one buy, sell or hold per agent per round, all of it trading the gap between price and a fair value signal. Systematic gap trading, not high frequency market making. 1️⃣ 𝗧𝗵𝗲𝘆 𝘁𝗵𝗶𝗻𝗸 𝗮𝗹𝗶𝗸𝗲. The stronger two models were, the more their decisions matched. Coming from rival companies barely helped. 2️⃣ 𝗪𝗶𝘁𝗵 𝗴𝗼𝗼𝗱 𝗶𝗻𝗳𝗼, 𝘁𝗵𝗮𝘁 𝗶𝘀 𝗴𝗿𝗲𝗮𝘁. More AI traders, better prices, every time. 3️⃣ 𝗪𝗶𝘁𝗵 𝗯𝗮𝗱 𝗶𝗻𝗳𝗼, 𝗶𝘁 𝗯𝗿𝗲𝗮𝗸𝘀. Feed them all the same misleading story and mispricing hits 𝟱 𝘁𝗼 𝟳 𝘁𝗶𝗺𝗲𝘀 what random traders produce. One model called the direction right 𝟰𝟮% of the time, 𝘸𝘰𝘳𝘴𝘦 𝘵𝘩𝘢𝘯 𝘢 𝘤𝘰𝘪𝘯 𝘵𝘰𝘴𝘴. 💡 Mistakes that differ cancel out. Mistakes everyone shares pile up. Testing one model at a time will never show you this, and switching vendors does not fix it. 🔎 𝗢𝗻𝗲 𝗰𝗮𝘃𝗲𝗮𝘁: it is a simplified simulation with a single asset and a known true price. Trust the mechanism, not the exact numbers. #AI #ArtificialIntelligence #LLM #MachineLearning #SystemicRisk #QuantitativeFinance #FinTech #AIAgents #FinancialMarkets #RiskManagement #AISafety #AIGovernance #Trading #AIResearch #MIT #DeepLearning #GenerativeAI #AlgorithmicTrading #Innovation #Technology