Check out how to use ChatGPT in financial analysis
AI productivity boost for your team. AI automation, consulting, training. Find out how to 3x your productivity.
4 studies about #AI lead the way to using #LLMs in financial analysis. I put together a sample prompt you can tune to fit your financial analysis needs. I mostly used it to get analysis of US publicly traded companies. The 10-Q, 10-K, earnings call transcripts, and other info. Analyzing financial reports seems to get pretty good results. If you have done any investing at all you know that analysts’ opinion is more or less a coin toss. Analysts are not interested in you making money but trading often. #ChatGPT doesn’t have that bias and it outperforms human financial analysts on average. But that’s not all, GPT-4 is also outperforming specialist AI systems. This is good news if you want to manage your own investment portfolio. Put 90% in an index fund and use 10% for your experiments in the casino called stock market. Another interesting study DELVES into REALM of creating narrative predictions. Basically, you make GPT tell you a story based on the information it has. You can also add additional information as a file or directly in the prompt. For example, when I experimented with stock prices, giving it a correct starting point improved the results. I used it for my own holdings of MSFT, GOOG, NVDA, and others. If you had to read those reports word for word it would take you several hours each. I fed the tool info from the past and asked it to come up with predictions for the period between its end of training data and now. The results were pretty good. Of course, not statistically significant. The study about getting the point out of bloated CEO talk reminded me of Isaac Asimov’s Foundation. In one situation they analyzed what a diplomat said during a multi-day visit. The result was that he said NOTHING. Now you can ask an LLM to do the same thing to the approximately 10k words they output during the quarterly earnings calls. In one example the result from AI is 4x shorter than what the CEO said. Finally, #GenAI is also helpful in uncovering corporate risks within financial documents. AI estimated political, climate, and AI-related risk based on earnings calls and their own information. You can get more details and all the links to the studies in the article. Link in the description. Also sign up for my newsletter to get all the important information about how to make AI useful in your work.