How to Use AI to Research Stocks (With Brian Feroldi)

0:00Fourth of July savings are happening now at the Home Depot with select appliances starting at$398. Plus, get free delivery on appliance purchases of$398 or more, no membership required. Upgrade your kitchen with a modern and sleek GE profile refrigerator featuring hands-free autofill for the perfect pour every time. And make laundry day easier with two-in-one washer-dryer combo innovation that completes laundry in about 90 minutes. Shop top brand appliances now at the Home Depot. Offer valid June 17th, July 8th, U.S. only C-Store online for details. on this episode of the Personal Finance Podcast, how to use AI to research stocks with Brian Feraldi.

0:54What's up, everybody, and welcome to the Personal Finance Podcast. I'm your host, Andrew, founder of MasterMoney.co. And today on the Personal Finance Podcast, we're going to be diving into how to research stocks by using AI. If you guys have any questions, make sure you join the Master Money newsletter by going to MasterMoney.co slash newsletter. And don't forget to follow us on Apple Podcasts, Spotify, YouTube, or your favorite podcast player. And if you want to help out the show, consider leaving a five-star rating and review on Apple Podcasts, Spotify, or again, your favorite podcast player.

1:28Now today, we're diving into something that could completely change the way you research and analyze stocks without replacing your judgment or turning you into a robot investor. Now, if you've ever stared at a 200-page 10K and thought, there has to be a better way. This episode is for you because AI won't make investment decisions for you, but it can make researching businesses dramatically faster, cleaner, and way more structured. And to break this down, I brought in the perfect person to talk about it, who is Brian Feraldi, a long-term investor, author, and educator known for making fundamental analysis simple and approachable.

2:02Now, Brian has been on the show many times. I think this is his fifth time on the show. And anytime we're talking about stock investing, we're bringing Brian on the show. Now, we're going to unpack how to actually use AI like a junior analyst, how to give it structure so it finds the signal, not noise, and how to demand citations so you know you can trust the data. Plus, we're going to talk about how to break down companies into business models, moats, financials, and risks without getting lost in a black box. And we'll even walk through a live demo of prompts that Brian uses in real life to break down companies step by step.

2:33The same kind of workflow that can help you analyze businesses faster and smarter. Now, if you want to watch as we do the live prompts, you can watch this on YouTube. Again, search my name, Andergen Cola, on YouTube, and you will find us there. And we're going to be doing live prompts and sharing the screen on YouTube. So that's a great place to watch this if you want to, but you can also listen along if you are listening to the audio version. So this is a great episode, an action-packed episode. So without further ado, let's welcome Brian to the Personal Finance Podcast. So Brian, welcome back to the Personal Finance Podcast.

3:06Andrew, awesome to be here. Thanks for the invite. I am really excited to have you here because I think today we're going to be going through a bunch of cool things that you can utilize AI when you are investing in stocks. and we're gonna do a bunch of research here. And we get a lot of questions about this. This is one thing I think most people are trying to figure out. A, how do I utilize AI when it comes to my finances? B, how do I utilize it when it comes to investing? And I think it's going to be a very powerful investing tool. It is already today, but then it's going to become even more powerful here in the near future.

3:33So we're going to just get right into some of the tactics that we want to dive into. And most of you have heard Brian on here before. Brian, this is probably your fifth time on here, I think. And every single time that we have you on here, a lot of people come away with a lot of tactical things that they can do right away, which is why I absolutely love having you here. So this is going to be really, really fun. So first, I want you to kind of talk through just your long-term mindset when it comes to investing that you learn from Buffett and Munger, because you and I align on this a lot. And we are long-term investors.

4:01We are investors who like to invest long-term. So can you kind of talk through your mindset and how you think about that? So I tried every investing style that exists out there, and I failed miserably on every other style. And it wasn't until I started to study the investing greats like Buffett, like Munger, like Peter Lynch, that I really found a style that fit my personality well and actually generated positive returns for me. So if you study the biggest investors, the best investors of all time, they all talk about the same fundamental principles. Think of stocks not as tickers, but as businesses.

4:35Only invest in businesses that you understand and you think have good long term growth prospects. Use valuation techniques to buy those companies below their intrinsic value and let the companies do the hard work. Before doing any of that style, which is my current investing style, I tried day trading. I tried penny stocks. I tried going for high dividend yield stocks. If I was just starting investing today, I guarantee you I would be on Robinhood trading crypto. I would be trading options. I would be doing all the things that new investors are doing today simply because they don't have the right mindset for actually building wealth in the market.

5:11So there's nothing wrong in my book with trying to create money in the short term through the market. I just don't know how to do it reliably. So my fallback is buy and hold great businesses for the long term because that's the style that works. I went through the same exact cycle where I was trading penny stocks. I day traded for six months full time, literally did not make any money. I lost money while I was doing it. For six months, I thought I could figure it out and just never did. I started to invest in just a bunch of different things and then realized, okay, long term is the way to go.

5:39Because every single time I try to figure out how to time the market or anything like that, I'm wrong. And so for now, we're going to go long term. And that's when I started to kind of go into individual stocks, index funds, and ETFs, those types of things. So that was a big, big difference maker for me. Now, AI helps a lot of people with a lot of different things right now. And specifically, it can really help you when it comes to researching specific things. So why do you believe AI can kind of help us accelerate our process when it comes to researching as an investor? Because this is going to be a tool that a lot of us can really help utilize.

6:11We're digging through 10Ks and all these different things. But instead, how does it actually help someone like an investor out there accelerate their process when they are researching stocks? So AI, by its basic design, is a wonderful resource for investors to use because at its core, AI is really fantastic about ingesting gargantuan sums of data and then using what it learned from ingesting that data to help users come up with insights from huge amounts of information. When you boil it down investing to its core or investing the way that I do, which is individual company analysis and then adding them to my portfolio, that's a perfect use case for what AI is built for.

6:50So with AI, you can take a long multi-hundred page annual report of a company, uploaded to AI. And if you prompt AI the right way, it can parse through all that information very quickly and help you to glean some insights from it. It can also go through conference calls, it can go through earnings reports, it can go through proxy statements, and it can explain to you in plain understandable English what you need to know about that company. So it dramatically lowers the skills that you need to get started with analyzing individual businesses, and it can act as a tutor slash buddy for figuring out what the terms mean.

7:25So it is a godsend for new investors that are trying to learn how to analyze companies. I think that's the most powerful thing is for most people out there, this can tremendously help you, especially if you're a brand new investor. This is going to help you so much more than what we used to have to do. We'd have to kind of read all these books and figure out what we need to do. I had to figure out, you know, how do I even look at this 10K and how do I read through all this stuff? And so this is going to help you tremendously when it comes to that research process for sure. Now, can you explain your junior analyst analogy and how that mindset helps investors utilize AI more effectively.

7:56Yeah. If you are going to be using AI to help you research companies and make decisions, it's really important that you approach AI with the appropriate mindset. And I stole this from somebody else, but they basically said, think of AI as a junior research assistant or think of them as an intern. If you were a fund manager and you have analysts working for you, you would not outsource all of the key critical information to that junior analyst, nor would you just give that junior analyst unlimited control over what it does. If you were going to be using an intern or an analyst the right way, you would give them a specific task, give them a specific set of instructions to follow.

8:34You would limit the number of resources that they could use, and you would have that analyst or junior analyst ask you questions to make sure you're getting the information that you want. As an example, if you've ever gone to ChatGPT or Claude or Grok and say, hey, is Amazon stock a buy? That is way too broad of a question to possibly ask an AI because it doesn't know what specific steps or instructions you want to follow. So it will do exactly what you tell it to do. So it might go out and look at Amazon's recent earnings report. It might look at sentiment. It might look at technicals. And none of that might be useful to your specific investing process.

9:12So if you don't constrain the AI into the instructions that you specifically want it to do, it's not going to give you anywhere close to the right analyst. The same way if an intern fresh out of college, you said, hey, is Amazon a buy? If it knew nothing about your investing philosophy or your investing process, it's going to go out, find information, and present you with information that is basically utterly useless to what you want. So if you have the right mindset of thinking of the AI as a junior analyst, that's really going to help you to come up with a prompt to give that analyst instructions to follow to make sure you get exactly the information that you want and none of the information that you don't.

9:49I think that's one of the most powerful mindset shifts that people need to have is you can't just take every single thing that it says, kind of utilize that information to take action. You have to utilize it as someone who is an advisor or a junior analyst in your life that can kind of help you make decisions as time goes on. Now, when do you think investors should kind of think through and utilize their own judgment? Because your judgment still matters in today's day and age. So when should they utilize their own judgment rather than AI? Yeah, you have to be the ultimate decision maker. After all, if you're going to make an investment, this is your money that's on the line.

10:19And if the AI makes a mistake in the analysis, it's going to cost them nothing. And if anybody here has been using AI, you've probably noticed that the AI gives you the wrong information or the analysis back. nothing related to investing. And then when you push back or correct the AI, at least for me, it always says, you're right. I totally overlooked that. Way to go. And it has a very positive bias to it. So I would never outsource all of my decision-making and thinking to AI. That would be a bad use of AI, especially with investing. Again, this is my money that's on the line. So I want to make sure what AI is doing is the task that it's most well-suited for.

10:55Go out, find information, ingest information, analyzing it using the rules that I said, and then present that information back to me so that I can make a more informed decision. That to me is using AI the right way. It is, and I think there's all these memes that are coming out now where I saw this commercial the other day where a kid was walking into a gas station and the gas station attendant asked him, hey, how's your day going? And he looked at his AI and said, my gas station attendant just said, how's your day going? What should I do next? And that's how some people use AI. They think of it and they use it as their only thoughts, especially with stuff like this.

11:25And you gotta make sure that you're still utilizing your own judgment because it's really, really important, especially as we start to go and AI keeps advancing. And this is going to be really, really important for a lot of people to be able to do that. So a lot of people also worry about AI hallucinations. And so how do you personally make sure that the information you're seeing is trustworthy and kind of vet that information? Yeah, this is a huge, huge problem. Again, if you've used AI for any decision making or research in any other aspect of your life, you've probably see it bring back information and you just intuitively know that's wrong or that's not right.

11:56and AI tools that are out there are unbelievably complex and sophisticated, but they can provide you with wrong information or they can make up facts and then present them to you. So that is a huge problem. In fact, when I was doing research on how to use AI the right way, I was talking with lots of investors that are in my community and this was the number one concern that they had before they would bring AI into the research process. They were basically like, how can I possibly trust the information that's coming back from AI given the hallucinations that happened? I do think there are steps that you can take to minimize the number of hallucinations that exist and to make sure that you can actually trust the information that comes back.

12:34But again, this is why you can't ask it broad, unbounded question like, is ABC stock a buy? You have to get much more granulated than that and you have to provide it with very detailed, highly specific prompts to make sure the information that you get back is trustworthy. And that's, I think, one of the most important things is kind of breaking it down and understanding how to prompt it properly. I know for a lot of times, we do a lot of really deep research with AI when it comes to just even looking at historic market data, or we'll look at statistics and kind of break down some really advanced statistics when it comes to people's finances.

13:02And when we do that, if I try to do one big, large prompt, it is something where I don't get the information back that I want. And so I have to break down the prompts and the research into much smaller chunks. So why for you is breaking research into smaller modular prompts so important? Yep. So with any complex task, again, researching a stock from scratch, at least for my process, there's lots of little micro tasks that are embedded in that question, such as what does the company do? What is its business and business model? Another task is what is the company's moat and competitive advantage?

13:34Another task is what are the company's financial statements? Another one is what are the company's growth prospects? Another one is what are the company's risks? What are the company's management teams? What's the company's valuations, et cetera, et cetera, et cetera. Again, if you ask a very broad question to analyze a business and it's going to have to go through each of those steps, if you don't break those steps down into miniature steps for the AI to tackle on, it could be reasoning up from faulty information. So if it gets the business and the business model wrong, how could it possibly have the moat and the growth potential right if it's literally looking at the wrong company or the wrong competitive advantage?

14:06So breaking complex tasks down, like researching a stock into smaller steps, and you can verify the information of those steps as you go, allows you to use building blocks that you can trust and building on top of that as your research up. So yeah, just an overall great thing to do with AI in general is to take big tasks, break them down into smaller micro tasks, verify the information of each of those tasks as you go. And that will give you much more confidence in the final output. And for those of you out there listening who just have not been prompting properly, you feel like you've been using AI and you use just, you know, a couple of lines and all of a sudden, you know, you're getting all this information back and it's not as detailed as you'd like it to be.

14:43If you break it into smaller chunks like what Brian's talking about here, this is going to make a big, big difference and just get way more granular and specific, that's going to help you tremendously. And if you develop processes like we're talking about today, that is going to change the way in the information that gets spit out to you every single time you have these conversations with AI. One thing I've learned from running multiple businesses is there's usually a gap between how you think work is getting done and how it's actually getting done. And we've had times where we thought we knew the bottleneck only to realize later the real issue was somewhere completely different.

15:17That's exactly what today's sponsor Scribe helps solve. Scribe is a workflow AI platform trusted by 94 % of the fortune 500 and Scribe optimized turns real workflow data into decisions your AI strategy actually needs automatically capturing how work happens across approved apps, surfacing inefficiency and opportunities with no interviews and no manual discovery and no extra work from your team, which is the key. And the live dashboard shows where work is happening, how much time it's taking, and where the biggest opportunities are right now. It automatically discovers workflows across multiple tools, then highlights your biggest inefficiencies and gives AI-powered recommendations with estimated time saving.

16:00You know I love saving time. It's also built with privacy in mind. It only runs across approved business applications, user-level data is anonymous by default, and sensitive information is automatically redacted and never leaves your firewall. To see, optimize in action, head describe.how slash PFP and mention PFP for a 30-day risk-free trial. That's S-C-R-I-B-E dot H-O-W slash P-F-P. So when it comes to this, so we're talking about ways to research, we're getting this information back. How do we kind of gauge the difference between our own judgment and what AI is telling us? How do we make decisions based on having two sides of the coin?

16:39Because we could have previously believed one thing, AI spits out some different information. So how do you kind of utilize your judgment and balance that with AI's information? Yeah. So there's a couple of core prompting techniques that I think are just good AI hygiene that investors can use to make sure that the information that they're getting back is as trustworthy as possible. And again, if you don't have full confidence and full trust in the process and information that you're getting back, doing this analysis will be utterly useless to you. It won't give you any advantage at all. Having conviction in the information and making decisions making on solid information is absolutely foundational.

17:14So there are three techniques that I think listeners should know. And if you do all three of these techniques, I think your confidence in the information that you're getting back would just be sky high. So one technique that I think everybody should do when using AI is to assign AI to be a specific role. As a really simple example, let's say you're a value investor and Warren Buffett is your investing hero. hero. One of the things you should put into the prompt is act as Warren Buffett. Act as Warren Buffett. If you give it that simple one sentence up front, what AI will do is go through its vast database of information that's using to making decisions, and it will prioritize information that is pulled from analysis on Warren Buffett.

17:55So it might use Berkshire Hathaway's annual meetings transcripts. It might use interviews that Warren Buffett has done. And immediately, The AI will know that you're looking at value investments. It might look at Warren Buffett's portfolio to pull information from. It'll know what the terms like moat and management mean. It'll know to emphasize factors like return on equity and competitive advantage. And that simple one-word phrase, act as Warren Buffett, will dramatically improve the quality of information that comes after that. And of course, you don't have to limit it to Warren Buffett. You can say, act as a value investor.

18:28You can say act as Joel Greenblatt. My favorite investor to follow is a guy named Terry Smith of Fundsmith. So I would say act as Terry Smith. If you're a short seller, you could say act as a forensic accounting. So assigning the AI a role up front, wow, that one step alone will dramatically increase the quality of information you get back. 100%. And I think that is one big thing that you could use in everyday life across the board. I've utilized AI and assigned it a role based on, you know, specific athletes. So one big thing I did over the course of the last year was trying to get in the best shape I possibly could.

19:01And so I looked across the board and I was looking at, you know, different athletes out there and what they are actually doing. And so if you can assign that role to a specific person, it will act in that way and it will help you kind of make decisions. And I think that's really, really important. So you also have some other core prompting techniques. So the first one is assigning a role, making sure we have that role in place. But what other techniques can you talk through in terms of how people can actually prompt AI to make sure they get the right answers back. So prompting technique number one, force it to assign a role.

19:30Technique number two is to force it to use citations and force it to look up information from resources that you trust, that you personally trust. Because if you're restricting the resources that it can use to create that information, you're building on a solid foundation. Now, for me, when researching companies, resources that I trust are very, very few and far between. I trust the SEC filings. I trust information that comes directly from the company. And then I trust a few other sources such as Morningstar's MoatScale. So when it's presenting with me with information, I force it to put a citation with a link to where it got that information from.

20:09So as I'm reading the results of the prompt, if I'm like, where did it get that information from or why is it saying this? If you click over to the citation, it will take you directly to the SEC filing or the company filing that that information came from so that you can confirm easily that the information that it's building on top of is correct. So again, if you limit the places that the AI can pull sources from and you force it to give you a citation of where it came up with that source from with a direct link, that dramatically increases the trust that I have and the information it's giving me.

20:40For sure. And I think these direct links have gotten so much better. I remember early on prompting AI, like, where'd you get that information? It would try to pull sources from just these random places. But once it kind of understands the core sources that you want to have in place, it'll remember that information. It'll make sure to pull from these places that you actually wanted to pull. For example, one of the places we always wanted to pull is all the Federal Reserve data that comes out when it comes to consumer spending and finances and those types of things. And so if you can kind of set it up in a way where it knows what sources you like, then that's going to be really, really important.

21:09And then you also talk about when we're prompting, we need to make sure we have the proper order and the proper steps in place. Can you kind to talk through how we set that up as well. Yep. So step three here is to give the AI a step-by-step instruction to follow to make sure that it is following the process that you would use to analyze a company in its own way. So for example, when I'm analyzing a business for the first time, the very first thing I want to know is what does this company do? What are its key products and services? Who are its customers and how frequently do they buy? What geographies does the company operate in?

21:42What are the revenue segments by product line or category? And then I also want to just know about the company. Do I think this company can raise prices in the future? What happens to this company during a recession? Are its product and services highly cyclical or are they recession proof? So a prompt that I built helps me do exactly that. So the prompt goes step-by-step through answering the specific questions that I want to know about any business in the order. And then it presents it back to me, answering it in the exact stepwise function that I have. So not only are the prompts that I've built, built out to answer steps in a stepwise function, but I use the prompts to analyze one part of a business and I give it specific instructions for how to do so.

22:22So if you have an investing checklist that you use, or again, if you wanted to analyze a stock like Warren Buffett, you can literally say to the AI, follow Warren Buffett's investing checklist, and it will know the process that it should use to research information. So if you tell it, do A, do B, do C, do D in that order and use citations, the trust that you will have in the information that comes back to you will be sky high. And this is why it's so important, I think, for a lot of people out there to start to build out a bank of prompts that they're going to be utilizing more frequently, because a lot of people, what they do is they throw it into chat GPT or whatever AI they are utilizing.