This paper investigates whether a large language model (LLM) can perform financial statement analysis (FSA) as effectively as professional human analysts. The authors use GPT4 to analyze standardized and anonymous financial statements to predict future earnings direction. Despite lacking narrative or industry-specific information, the LLM outperforms financial analysts in predicting earnings changes, especially in challenging situations. The LLM's prediction accuracy is comparable to that of state-of-the-art machine learning models, and it generates useful narrative insights about a company's future performance. Trading strategies based on GPT's predictions yield higher Sharpe ratios and alphas compared to strategies based on other models. The study suggests that LLMs may play a central role in decision-making in financial analysis. The authors also explore the LLM's performance relative to specialized ML applications and find that while the LLM's accuracy is similar to that of narrow ML models, it excels in tasks requiring human-like reasoning and intuition. The paper contributes to the literature on fundamental analysis and the relative advantages of humans versus AI in financial markets.This paper investigates whether a large language model (LLM) can perform financial statement analysis (FSA) as effectively as professional human analysts. The authors use GPT4 to analyze standardized and anonymous financial statements to predict future earnings direction. Despite lacking narrative or industry-specific information, the LLM outperforms financial analysts in predicting earnings changes, especially in challenging situations. The LLM's prediction accuracy is comparable to that of state-of-the-art machine learning models, and it generates useful narrative insights about a company's future performance. Trading strategies based on GPT's predictions yield higher Sharpe ratios and alphas compared to strategies based on other models. The study suggests that LLMs may play a central role in decision-making in financial analysis. The authors also explore the LLM's performance relative to specialized ML applications and find that while the LLM's accuracy is similar to that of narrow ML models, it excels in tasks requiring human-like reasoning and intuition. The paper contributes to the literature on fundamental analysis and the relative advantages of humans versus AI in financial markets.