Inside the Pod: How to Use AI to Be a Smarter Investor

Thank you to everyone who is watching or listening to my podcast, How Do You Use ChatGPT? If you want to see a collection of all of the prompts and responses in one place, Every contributor Rhea Purohit is breaking them down for you to replicate. Let us know in the comments if you find these guides useful. —Dan Shipper

The best investment opportunities are found in moments of perceived crisis. But not every crisis will be a good investment.

In fact, identifying which moments of crisis are promising opportunities is an art. It’s the stuff that professional Wall Street traders grapple with every day—and something amateurs like us pore over, too, determined to beat the market.

In this episode of How Do You Use ChatGPT?, Dan Shipper and IA Ventures managing partner Jesse Beyroutey tell us how they outplayed Wall Street. Dan and Jesse invested in Nvidia in 2019 when the shares were trading at $33. They’re worth nearly $900 a piece today. It was the best trade of their lives. In this interview, Dan and Jesse walk us through how they used Google’s powerful LLM Gemini Pro 1.5 to try and top that.

To get them started, Dan opened an account on investment platform Robinhood and put in $1,000 as the principal amount. With 90 minutes on the clock, Dan and Jesse were off to the races.

This interview is a masterclass in how to use AI to refine your own investment thesis and make smarter financial decisions. I came away inspired—and more than a little tempted to test my luck and day trade.

First, we’ll give you Dan and Jesse’s prompts, followed by screenshots from Gemini and ChatGPT. My comments are peppered in using italics.

A page from “Beyroutey x Shipper Investments,” a Notion page which records why they thought Nvidia would be a good investment in 2019—even though the company’s stock price was down. Jesse says they fancied themselves as “distressed asset investors.”

Dan and Jesse invested in Nvidia because they believed that the company was temporarily underperforming on account of identifiable reasons recorded in the screenshot above. With the clock ticking, Jesse uses the stock analytics platform TradingView to apply search parameters that might mimic this criteria. He adjusts the filters to find companies with a market capitalization between $2 billion and $100 billion, a positive gross profit, and a downturn in performance over the prior six months.

Dan and Jesse: Hey, I’m trying to pick a stock, I want to find a tenbagger. Specifically, I'm trying to replicate a trade I did in 2019 where I bought Nvidia. It was down temporarily because of the trade war, and now it's a $2 trillion company. Here's a list of potential starting points for trades. Does anything pop out to you?

Dan and Jesse: How should I construct a screener to identify companies that might be artificially depressed in a stock screener?

Dan and Jesse: Okay make a decision. Do the best you can.

Dan and Jesse: Pretend you're Warren Buffett and I've come to you to make a final decision on a trading strategy.

Dan and Jesse: If you had to cluster these stocks, how would you do it and why?

Here’s a slice of what Gemini generated.

Dan and Jesse: Please cluster the stocks by what the company does. We’d like to analyze it on the basis of the nature of the company and its position in the industry.

Dan and Jesse: Which of these businesses are most likely to be trading poorly because consumers dislike the brand?

Dan and Jesse: How hard would it be for GoGo to produce hardware that would allow it to upgrade its current services in planes to improve service quality using the latest in connectivity like Starlink?

As they wait for Gemini to respond, Jesse notes that based on their analysis so far, GoGo resembles their worst investment more than it does their Nvidia trade. He says that they assumed that the stock price of their worst investment was low because the company had just been sued. The investment thesis was grounded in the low price being a “temporary dislocation” as opposed to a “wave of positive momentum” from being well-positioned to adopt a new technology (like Nvidia).

Dan and Jesse: To what extent is management already talking about upgrading to new technology?

Dan and Jesse: To what extent would that upcoming change already be priced into their stock and how would we figure that out?

This hiccup also prompts Jesse to pause and reflect on the original investment thesis of finding stocks trading down for a clear reason. He thinks that instead of using parameters on a stock screener to find companies that are underperforming, they should search for stocks trading down because of “exogenous factors”—like inflation or industry trends. Dan and Jesse decide to try the AI search engine Perplexity to answer this question.

Dan and Jesse: What are some stocks trading down because of exogenous factors?

Dan and Jesse: What energy companies control a key bottleneck in the value chain and are positioned to ride a new technology wave over the next five years?

ChatGPT:

Dan and Jesse: Give us some public stocks that fit each of these categories. 

As they go down the list of companies, Jesse revisits the investment thesis for the Nvidia trade and documents the characteristics that made it successful:

• Stock trading down because of exogenous factors
• Stands to benefit from a technology wave

While they have included both these parameters in their search for the next tenbagger, Jesse realizes that there was another factor that contributed to Nvidia’s success:

• An extremely scalable business model

Dan and Jesse return to the same chat window where Gemini had compiled the list of publicly traded energy companies. 

Dan and Jesse: We’re looking for a company that has a particularly scalable business model, so that as the technology wave—for example: solar—comes to bear in the coming years, this company stands to benefit, but also scale massively. Which public companies in energy also have a highly scalable business model? 

Jesse uses the example of Nvidia and Meta to illustrate how companies can have highly scalable business models in different ways:

• Nvidia is “fabulous as a manufacturer” and can buy more capacity from the chip manufacturer TSMC when it experiences an increase in demand.
• Meta, on the other hand, doesn’t need to spend much more on infrastructure or any other thing when advertising spend grows. 

They try prompting both Gemini and ChatGPT again, this time with more detailed instructions. 

Dan and Jesse: We’re trying to find a company whose business model is scalable in that it can grow 100 times in a year and not run into any key bottleneck. That characteristic is extremely rare and more common in the digital world, but we're looking for things outside of [the] internet and software. In particular because energy seems like a big opportunity, we're wondering if any companies in the sector have this sort of characteristic, and sit in a bottleneck of the value chain that allows the company to capture the lion's share of new opportunity as it comes up in a tailwinded market.

Gemini:

Dan and Jesse aren’t particularly impressed with the results generated by ChatGPT or Gemini (especially since it had already named the four companies in its previous responses). They decide to start over with a new prompt that goes back to their fundamental investment thesis.

Dan and Jesse: We're looking to find public stocks that have the following characteristics: are suffering some kind of exogenous shock to their business, sit in a highly valuable bottleneck in the value chain, and have a highly scalable business model such that they could scale 100 times in a year relatively easily. List them.

Here’s part of Gemini and ChatGPT’s response. 

Gemini:

Dan and Jesse notice that Gemini listed stocks under each of the three categories, instead of a consolidated list of stocks that have all three characteristics. Either way, since both Gemini and ChatGPT found very well-known stocks, they ask the models to dig deeper.

Dan and Jesse: Can you dig deeper and find some less well-known stocks?

Here’s a slice of what ChatGPT generated.

ChatGPT:

Dan and Jesse: Can you give me a list of all of the companies that went public via SPACs in 2020 and 2021?

ChatGPT:

Dan and Jesse think that the scandal might be the “exogenous shock” they’ve been looking for, and that genetic data is going to be more valuable in the future. However, they can’t help but wonder if the company is going to go bankrupt.

Dan and Jesse: Is 23andMe going to go bankrupt? Make your best guess.

Dan and Jesse: What is Virgin Galactic up to? Browse.

With time running out, Jesse and Dan revisit 23andMe. Even though they don’t think it will be the next tenbagger, keeping with the premise of the episode, Jesse and Dan decide to invest in the stock. Dan makes the purchase on the brokerage app Robinhood—and just like that, they are officially investors in 23andMe.

As the episode concludes, Jesse reflects on their use of LLMs in the process of identifying the right stock. He says that while Gemini’s large context window is incredibly powerful, it can only be leveraged if there is enough content to upload to it. Jesse thinks that it’s difficult to find large amounts of usable content on the internet—a roadblock they ran into while trying to upload investor reports to the LLMs. He thinks solving for this problem, and connecting a real-time data source to keep these models updated, are new business opportunities unlocked by large language models.

Jesse and Dan remark that their Nvidia investment thesis is strong, but they weren’t able to find a company that fit their criteria at the moment. However, they believe that in the coming years, they are likely to come across the right opportunity…or maybe sooner—like one week later. 

Neither of them is particularly happy with their 23andMe investment, and when Dan notices that they’ve actually made a small gain on their investment a week later, he decides they should take the win, exit their position in 23andMe, and make a fresh investment.

While Jesse notes that Google’s year-over-year performance is still in the green, the stock is definitely trading below where it should be because of a variety of factors, including poor reviews of Gemini, lower-than-expected reported advertising revenue, and an unusually high capital expenditure. He thinks that these circumstances meet the “exogenous shock” component of the thesis. As for the other two parts of the investment strategy, Jesse and Dan conclude that Google has a uniquely scalable business model that is well-positioned to benefit from the AI wave.

If you have frameworks or models about how the world works, AI is going to be a potent force in helping you leverage your thinking—and I think this episode is a perfect example of this in action. In any case, it looks like we need Dan and Jesse to record the sequel to this episode in five years, so we can find out if their strategy worked.