Alphawatch

Tomorrow’s top companies will be run largely by AI and a few smart people who know how to apply that AI at scale! The best thing you can do right now is to learn to apply AI and start automating processes!

What does everyone think of the LLM outperforms Financial Analyst paper draft from U of Chicago? It's quite long, so here is the AI summary: The study "Financial Statement Analysis with Large Language Models" investigates the performance of large language models (LLMs) in financial statement analysis compared to professional human analysts. Here are the key findings: 1/ Performance Comparison: LLMs, specifically GPT-4, outperformed human analysts in predicting future earnings changes. Analysts achieved an accuracy of 53% while GPT-4 achieved 60% using chain-of-thought (CoT) prompting. 2/ Research Design: The study used standardized and anonymized financial statements to ensure the LLMs did not rely on prior knowledge or memory. The LLMs analyzed the balance sheet and income statement without narrative context. 3/ Benchmarking:GPT-4’s performance was compared to state-of-the-art machine learning models (ML) such as artificial neural networks (ANN) and logistic regression. GPT-4's accuracy (60.31%) was comparable to ANN (60.45%), indicating that general-purpose LLMs can match specialized models. 4/ Complementarity: The study explored the complementarities between LLMs and human analysts. LLMs provided incremental insights, especially when human forecasts were prone to biases or inefficiencies. 5/ Sources of Predictive Ability:The study ruled out the possibility of LLMs' performance being due to memory by anonymizing data and testing with out-of-sample data from 2023. LLMs generated useful narrative insights based on financial ratios and trends. 6/ Confidence and Magnitude:LLMs performed better when they reported higher confidence in their predictions. Predictions of large changes were more accurate than those of small changes. 7/ Generalizability:The findings were robust across different versions of LLMs, including GPT-4 and Gemini Pro by Google, with GPT-4 showing the best performance. 8/ Economic Usefulness:Trading strategies based on GPT-4’s predictions yielded higher Sharpe ratios and alphas compared to other models, indicating significant economic value. In summary, LLMs like GPT-4 can analyze financial statements and predict future earnings with high accuracy, potentially transforming the role of financial analysts and decision-making processes in financial markets.

From our preliminary testing, GPT-4o is maybe 5-10% better than Claude Opus, which is huge given the 3x cost difference.

We’re building AI workflows for investors. Here’s how it works: → Open your data room → Apply custom workflows → Analyze 1M+ files in seconds Transform any investor portal into a fully integrated workspace. AGI is coming. Will you be left behind? Or lead the way?

It’s official- The first human profession to be replaced by AI is an audio book reader! AI generated audio books are indistinguishable from human generated books

OpenAI just launched Voice Engine, It uses text input and a single 15-second audio sample to generate natural-sounding speech that closely resembles the original speaker. Reference and Generated audio is very close and hard to differentiate. More details in 🧵

Today, we're announcing Claude 3, our next generation of AI models. The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.

🚨 BREAKING: Nvidia just released Chat with RTX, an AI chatbot that runs locally on your PC. It can summarize or search documents across your PC's files and even YouTube videos and playlists. The chatbot runs locally, meaning results are fast, you can use it without the internet, and the user's data stays private. New day, new chatbot. Let's go.