When I’ve used ChatGPT to analyze stocks in the past, I’ve found that if you ask it the right questions, you can get some pretty impressive answers that might make your investing life a lot easier. So, naturally, when the latest and greatest version of the AI chatbot came out, I decided to see what it could do. Here are six smart ways that you can use GPT-4o to analyze a stock.
1. Gain a high-level understanding of a company
When you buy stocks, you become a part owner of the company, no matter how many (or how few) shares you hold. So treat each stock purchase as if you’re actually buying the whole company. And that means making sure you understand what the firm does and how it makes its money. Now, you can get that by reading annual reports, investor presentations, and so on. But you can also use GPT-4o to get an initial, high-level understanding. For example, I typed in, “Explain to me Tesla's exact business model”, and I got the answer below. (Note: because of the huge amount of information GPT-4o spits out, the images here are just snippets of the bot’s output. To see the full responses, click on the images.)
That’s a very detailed summary, telling me what Tesla sells and how (which is equally important). If you want to take it a step further, you can use GPT-4o to figure out a company’s economic moat – that is, its ability to maintain its edge over the competition, protecting its market share and profit over the long term. Legendary investor Warren Buffett always checks out a company’s moat before he invests. So I asked GPT-4o, “What are Tesla’s economic moats?”, and here’s what it told me:
Again, not a bad summary. You can get more creative too, asking specific questions to help you better understand a company. For example, I asked, “How easy is it for Tesla to scale?”, and I got a thoughtful response that discussed factors that help and hinder Tesla on that front, along with what the company is doing to expand.
2. Perform a SWOT analysis
You can also use GPT-4o to do a SWOT analysis – a fact-based, data-driven evaluation of a firm’s strengths, weaknesses, opportunities, and threats. This is a very useful exercise, but it can be quite time-consuming to do in real life. I prompted, “Write me a SWOT analysis on Tesla”, and GPT-4o’s response was a great start. Here’s what it said:
3. Summarize earnings calls
Earnings calls are a key source of valuable information for investors. During these hour-long conversations every quarter, a company’s top brass goes over the firm’s recent performance, discusses the outlook for the business and industry, answers questions from analysts, and more. But it could take ages to listen to the calls for every stock you’re invested or interested in. So why not ask GPT-4o to summarize them for you? Here’s how I did that.
First, I used Google to find the transcript of the specific earnings call I was interested in, searching “Tesla Q1 2024 earnings call transcript”, and that gave me a few links with the full transcript, including the management team’s prepared remarks and the Q&A session with analysts. I copied all the text and pasted it into GPT-4o, and then wrote this prompt at the very top (i.e. before the transcript): “Here is the earnings call transcript for Tesla. Please summarize the key financial results, operational highlights, future guidance, and main points from the Q&A session.” Here’s its response:
Useful, no? This is personally one of my favorite ways to use GPT-4o, but I did see some potential drawbacks. First, you might not always be able to find a particular earnings call transcript online (well, not for free at least). Second, actually listening to an earnings call allows you to gauge the management team’s tone, which can offer some valuable clues about whether they seem confident, nervous, evasive, and so on. That’s something that can’t be captured in transcripts. Third, the summary is sure to miss some important details.
4. Evaluate a company’s ESG credentials
Investing is about more than just returns: you can choose opportunities based on the effects that companies have on the world around you. There are several ways to go about it, but one of the most popular methods is via environmental, social, and governance (ESG) investing. And there’s evidence suggesting that firms with good ESG credentials outperform their peers over the long term. So with that in mind, you could use GPT-4o to see how good (or bad) a firm looks through an ESG lens.
For example, I typed, “Evaluate Tesla’s ESG credentials”, and I got a pretty darn good summary:
5. Generate code to backtest buy and sell signals
This is a more sophisticated way to use GPT-4o and one that’ll appeal to the coders out there. Say you wanted to see whether it’s a good time to buy a stock based on some technical analysis indicator – for example, the relative strength index (RSI). Now, remember, the RSI assigns a number between zero and 100 to an asset: below 30 means it’s “oversold”, potentially offering up an attractive entry point. (There’s more on the RSI here.)
Now imagine you wanted to see if it’s historically been profitable, on average, to buy a specific stock whenever its RSI dipped below 30. A simple question, sure, but one that’s quite hard to answer. The best way to do it though is via coding, but that’s a long and complicated process that requires some technical knowledge. Enter GPT-4o, which can write the code for you. I typed, “Write me code to see how Tesla's stock price performs one month after its RSI dips below 30”, and it did (here’s some of that code):
It even gave me a neat summary at the end, explaining each main step of the code. Not too shabby! This is just one of hundreds of examples of technical analysis things you could test out. As another example, you could ask GPT-4o to write code to see whether it’s historically been profitable to buy a stock when its 20-day simple moving average (SMA) crosses above its 200-day SMA.
6. Identify key risks
No investment analysis is complete without properly considering risk. The best investors out there are the ones who not only identify promising stocks, but also pay close attention to what could go wrong and hit the exits accordingly. If you’re not sure what the key risks are for a particular company, just ask GPT-4o for some initial ideas.
For example, I asked GPT-4o, “What are the key risks associated with investing in Tesla?”, and it identified eight main ones. That’s a great starting point, allowing me to pick the ones I think are most important to investigate further.
Looks good, but what are GPT-4o’s limitations?
Needless to say, GPT-4o isn’t perfect – neither is any AI technology at the moment. And there are some important limitations to be aware of when using it to analyze a stock.
First, it won’t be able to answer the questions we went through for every company – it’s going to struggle with smaller, lesser-known ones. I cherry-picked Tesla for all my examples specifically because it’s so well-known.
Second, as of the time of writing, GPT-4o’s training data goes up to December 2023. That means its answers can sometimes miss recent information if it’s solely relying on its in-built knowledge. Now, the AI can access the internet to give more up-to-date answers, but that means that the quality of those responses will be dependent on how good the underlying websites are.
Third, GPT-4o can sometimes spit out answers that are incorrect or entirely made up – a phenomenon called “AI hallucination”.
But despite the drawbacks, you can see how GPT-4o can help you more easily and efficiently analyze a stock. So while it should never be used in isolation, it can be an exceptionally handy tool that you can leverage to make your investing life easier. As the saying goes, work smarter, not harder…