Beyond The Horizon Can ChatGPT and DeepSeek Predict the Stock Market and Macroeconomy? Large Language Models (LLMs) like ChatGPT and DeepSeek can they really predict financial markets? This is not just about curiosity—everyone wants to make better decisions and generate returns. So, can we leverage the latest AI tools to rethink how we approach market analysis? Based on recent research analyzing 84,535 front-page news headlines and alerts from the Wall Street Journal (1996–2022), here are some interesting findings: 1.ChatGPT captures economic signals effectively: It shows a significant ability to predict stock market returns using headline data, going beyond just surface-level sentiment. 2.Good news has predictive power: The ratio of positive to negative headlines provides signals for market trends up to six months ahead. 3.DeepSeek captures sentiment—but with limits: While DeepSeek accurately reflects current market sentiment, it does not predict future returns. 4.Human bias vs AI analysis: Consistent with Epstein & Schneider’s (2008) theory of ambiguity aversion, investors react faster to bad news than to good news. Interestingly, ChatGPT excels at identifying undervalued positive signals—picking up on good news that human investors often overlook. 5. AI can extract signals that humans miss: ChatGPT’s emergent abilities enable it to identify patterns and insights that may be invisible to traditional analysts. This shows the potential of AI in financial decision-making. We’re just at the beginning of understanding how AI can reshape market analysis.