Paper vs. Profit
We read the papers. We judge them. We tell you if they will make you money. Brutal AI dissects academic finance research with one question: would this actually work in practice?
Three Korean researchers built a Random Forest model that trades individual KOSPI/KOSDAQ stocks intraday using dollar bars, triple barrier labeling, and a meta-model. 57.4 percent accuracy. Sharpe 2.77. Costs included. Quant fund strategy, not Robinhood strategy.
Soongsil University researcher reverse-engineered CNN's Fear & Greed Index, broke it into 7 components, and threw every machine learning model at it. The Stock Price Strength sub-indicator dominated. Direction prediction beat magnitude prediction. R-squared stayed under 0.20.
Academic research interpretation only. Not investment advice. Always verify findings against the original paper.