I Fed an AI Quant Agent Pure Noise. It Still Found ‘Alpha’ — Every Single Time.
中文摘要
研究显示，即使面对纯噪声，AI量化智能体也能挖掘出“Alpha”，表明LLM因子挖掘循环存在虚假发现的风险。
English Summary
An experiment shows LLM-based quant agents find "alpha" even in pure noise, demonstrating the risk of false discoveries in AI factor-mining loops.
Original Excerpt
A pre-registered test of what LLM factor-mining loops say when there is nothing to find. Continue reading on Medium »