Growth Investing

communities_1.jpg
 View Only
Expand all | Collapse all

Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

  • 1.  Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-21-2025 18:26

    The article highlights how large language models (LLMs) like ChatGPT are increasingly being explored for stock valuation analysis-especially for their ability to summarize qualitative data from earning calls, news, and filings. While these tools are still evolving, they offer intriguing potential to complement traditional valuation models and surface insights faster than manual review.

    Have you experimented with using AI tools like ChatGPT or others to help research stocks or interpret company fundamentals? Do you see them as useful for streamlining analysis-or do you still prefer hands-on review of financial statements and commentary?

    Let's explore how AI fits into our investment toolkit. If you haven't read it yet, the article explains both the promise and the limitations of LLMs in equity analysis: Using Large Language Models for Stock Valuation Analysis (July 2025, by Cynthia McLaughlin). 



    ------------------------------
    Lauren Ryoo
    AAII Marketing Intern
    Chicago, IL
    ------------------------------


  • 2.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-22-2025 12:21

    Maybe I'm a dinosaur, but I'll stick to doing my own analyses, and I'll use the "real" data to perform those analyses. LLMs have not been perfected. They lie, they get things wrong, and they have built-in biases. I won't say "never," but I do not foresee ever using an LLM to analyze investments. 



    ------------------------------
    Rob Adams
    ------------------------------



  • 3.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-22-2025 13:24
    Edited by BARRY JOHNSON 07-22-2025 13:31

    Robert, you are not a dinosaur. You are a realist in a hype-dominated world.

    That may make you out of step with the general population of your era -- a prolepsis or a solecism -- but so was Socrates. Socrates was a polarizing figure in Athenian society. In 399 BC, he was accused of impiety (worshipping false gods like "risk"?) and corrupting the youth (like other members?). 

    After a 1-day trial he was sentenced to death. He spent his last days in prison refusing offers to help him escape.

    I signed up to try out the new AAII Sentiment Investor offering because I am inquisitive (also a Socratic trait along with asking too many probing questions) about how this investing "ecosystem" works.

    Regards



    ------------------------------
    BARRY JOHNSON
    ------------------------------



  • 4.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-23-2025 11:16

    Barry, your kind words are appreciated, as always. I'm all for using tools that make analysis easier, but so far, I don't see AI doing that for me. AI is at its core a computer program created by humans with human input. Humans apply "guardrails" with all their inherent biases and flaws attached. When the first AI program generates Buffet-sized wealth, I might begin to see it differently. For now, I'm doing just fine with my stodgy old methods.

    You're right to apply a Socratic dialogue to AAII's new thingy. I'm pretty jaded on all of AAII's revenue-generating offerings, but in particular, I'm totally uninterested in any measure of "investor sentiment" other than market prices. 



    ------------------------------
    Rob Adams
    ------------------------------



  • 5.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-24-2025 17:02
    Edited by Chris Parks 07-24-2025 17:07

    I used Microsoft Copilot to create a summary of a company's annual report. Copilot created the summary in a few seconds. The summary saved me time, not having to read the entire report.

    I have all the AI articles in the AAII magazine. They are an excellent resource for educating investors on how to use AI. AI will be another powerful tool for individual investors.



    ------------------------------
    Chris Parks
    ------------------------------



  • 6.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-24-2025 18:41
    Edited by BARRY JOHNSON 07-25-2025 08:59

    Lauren,

    Your discussion premise is ... 

    chatbots "are increasingly being explored for stock valuation analysis especially for their ability to summarize qualitative data from earning calls, news, and filings."

    Since chatbots' only skill is summarizing (is it more accurate to call this integrating process "homogenizing"?) documents (chosen using preset algorithms that embed sampling and weighting criteria with a broad reach ("that's what large language model" means) and then algorithmically infers using an unknown set of assumptions),

    -------------------------------------------------------------------------

    the key question is

    what do I GAIN by saving myself a few minutes (let's say it's 3-4 hours) by NOT reading the original source document financial statements or a prospectus? 

    --------------------------------------------------------------------------

    Note: I find this as careless as asking the clerk who printed the closing mortgage documents you are not reading but are signing and nudging your spouse to sign too. Saying you did not read the source documents is nor a viable legal defense ... for all the reasons you KNOW.  Same with buying other high ticket, high risk transactions.

    --------------------------------------------------------------------------

    The perceived convenience may not be worth the increased level of RISK you introduce into the decision-making process that you make based on the representations in the source documents

    Buying a stock means you are becoming an owner. You should act like an owner

    ... just like you are when you buy a home, a car, or anything that has the potential to increase OR decrease your wealth accumulation outcomes. You need to take the time to take a test drive, kick the tires, etc.

    I don't EVER invest in stocks/funds that I expect to produce losses.

    -------------------------------------------------------------------------

    Buying stocks and bonds is serious business. I / you have a fiduciary responsibility (in the strictest  literal definition) and ... a huge ego at stake. I /you owe to my/your family and heirs who will base a large part of their memory of me on how well I do this simple task. Yeah, I know you think I am crass, but I have been to enough funerals to KNOW better. 

    ------------------------------------------------------------------------

    Saving 3-4 hours NOT reading an investment document is about the same amount of time it takes to play one round of golf. I love golf, but I know it takes time to perform the task. If you don't play golf, it looks so easy, but there are so many ways you can make small mistakes that produce bad results ... and penalties. I know of no way to explain the discipline and practice it instills and the level of skill it takes to perform well and "profit from the "skills" the "ancient" game brings to building practice, strategy, planning, patience, discipline, and practiced execution.  

    --------------------------------------------------------------------------------

    FASB financial statements, SEC filings, and a broker prospectuses prepared by multiple, nested, specialized  intermediaries -- all PAID by a "counter" party (who is not your friend -- to represent THEIR interest OVER yours -- already interjects multiple layers of opacity buried in several dozen pages of legalese that mediate the flow of cognitive sensors between the actual data you seek to parse and your brain's ability to process it.

    ---------------------------------------------------------------------------------

    Using a commercial chatbot deployed through a commercial operating system incorporating a large number subroutines/applets that introduce several additional layers that only increase the "density of the fog" that these nice folks require that you to accept as "the complete and legally-binding" documents that collectively focuses on just only one thing that matters at this moment -- you are relying on ALL the material representations these nested products produce ... at  your own risk.

    Ask yourself WHY all this legal density is there? 

    Notice you are provided the "convenience" of being asked to sign an acceptance UNREAD with a link to the official agreement. 

    If you want to grasp the level of disparity of interest here, just send THEM a document that says "This company, and all the organizations its relied on to provide this service, agree to treat you fairly," Then call and ask what the status is. The transcript of your call WILL be posted as the "Dumb Customer "Giggle of the Week.""

    ------------------------------------------------------------------------------

    Make no mistake. Every financial document you use to educate yourself about any financial offering is already "sanitized" by accountants and legal advisers.

    About 20 years ago I was reading a prospectus for some fund I someone recommended. I noticed that the pro and cons seemed to cancel out, so I highlighted all the "return/gain" statements in green and all the "risk/lose" money statements in red. And ... there were 52% gains and 48% losses/risks.

    Do you think that was an accident? You will flunk the licensing test.

    Look up "Caveat Emptor." That original Latin phrase has been around for a few years for a reason.

    Regards,

    ------------------------------
    BARRY JOHNSON
    ------------------------------



  • 7.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-25-2025 01:57

    I agree with Barry. The devil is in the details in analyzing any company. A summary is compiled based on whatever the summarizer thinks is important, and what the AI program thinks is important is likely to be quite different from what I think is important.

    By actually examining financial reports for decades, I've learned what is important (to me) and what isn't. Also, I learned long ago not to simply read them cover to cover. When I pick up a report, there are several key things I want to look at first, and those things lead me to other things I want to see, and so on. I can usually get through an annual report in 20-30 minutes---much less if I find a big red flag.

    I will confess that I frequently use compilations of data to make my analyses more efficient. My favorite is to see 10 years of financials side by side. But I always want to see the raw data instead of relying entirely on someone else's compilation. Even compilations have errors and treat numbers differently than I do. For example, do the reported earnings take "extraordinary events" into account? (I HATE "extraordinary event" reporting! I'd rather they just report the numbers as they fall and then explain any unusual influences in text or footnotes.)

    I suppose I should hope that most other investors will come to rely on AI, so AI can lead all of them in a herd this way and that. It would thus create more opportunities for crusty, old-fashioned individualists like me.



    ------------------------------
    Rob Adams
    ------------------------------



  • 8.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 07-30-2025 18:01
    Edited by Stephen Croce 07-30-2025 18:07

    "By actually examining financial reports for decades, I've learned what is important (to me) and what isn't."
    " I can usually get through an annual report in 20-30 minutes---much less if I find a big red flag."

    I might prompt LLM something like this: (oversimplified)
    Q:  Why types of things would a savvy investor consider to be a  "big red flag: when analyzing a company in the XYZ Sector
    A:  blah blah blah

    Q:  Here is the report.  Do you find any evidence of those "big red flags" you gave me in your last answer that warrant closer inspection?
    A:    Yes?  Please cite the sections including page number.  I read it myself do a few sanity checks.

    As good as decades of experience?  No.  But not a bad start.  Gets me a lot farther in 10 minutes that I ever could have by random googling or trying to scan endless articles.  Even if the information is somewhat flawed,  I know that going in.  Internet research was always flawed, and it still is - and the problem for decades has been not lack of information, but inundation of it and trying to separate signal to noise.   Assuming I am going to be looking at flawed information  anyway I'd much rather get it immediately in a few sentences and in natural language form.   Way better than nothing at all.   Way better than analysis paralysis and maybe I cut short at least a part of my learning cycle deficit by a few decades    If I want detail and accuracy, that's a different process that comes after having a general sensibility.



    ------------------------------
    Stephen Croce
    ------------------------------



  • 9.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-08-2025 19:58

    I just read a relevant article from Kiplinger's on this subject. It is entitled "Google's AI Overview Is Wrong About Life Insurance 57% of the Time, Says Study." Notable quotes from the article include:

    "'AI Overviews offer convenience, but their current accuracy for complex topics like life insurance and Medicare is unreliable,' said Choice Mutual CEO Anthony Martin." 

    "In addition to reaffirming the importance of factchecking AI, what this study also shows is how easy it can be for an AI response to fool even an intelligent user. These responses look thorough, containing specific details and precise terminology that can make the answers feel complete if you don't happen to have industry expertise or firsthand knowledge of the topic you're searching."

    "In many cases, it's not even immediately clear which aspects of a response might be wrong or warrant further research. According to Martin, this 'directly contributes to consumers making poor insurance decisions based on the false information provided by Google's AI answers.'"



    ------------------------------
    Rob Adams
    ------------------------------



  • 10.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-09-2025 08:46
    Edited by BARRY JOHNSON 08-09-2025 08:53

    Dr. Bob, thanks for sharing this article.

    You have turned the lights on to help us see a growing issue that needs more of our attention - the reliability of chat bot "information."

    #1 Last week (I have to go find it and post the link), a WSJ article chronicled how 5 or so AI chat bot provided incorrect responses to a series of basic decisions and choices for setting up a savings and investment plan.

    #2 In 2025 I have read several articles that had surveyed financial advisors (from MORN and SCHW) about what capabilities they most wanted most in using AI. #1 was to let them spend more time "building the business" -- I.e. sales and leaving portfolio management and written communications to the bots.

    I would ask any FA this question:

    What are you doing that is more important than managing my investment?

    If you don't make money for me, how do you expect to get more customers?

    I would want comparative data on FA performance since I am 100% sure that ...

    50% of FAs graduated in the bottom 50% of their FA licensing class. 

    It's a coin flip that you will get one of them.

    I have seen more but I only surveyed the headline. I'll post more.

    I apologize to all the AAII members who are licensed lifelong FAs, but my somatic markers -- anxiety, nausea, disgust -- (see Antonio Damasio's research -  Descartes' Error: Emotion, Reason and the Human Brain. Antonio R Damasio [1994]. Random House. ISBN 978-1-4070-7206-7. ) that tell my brain not to trust the motivations of anyone who solicits to manage my money -- banks, credit cards, investment industry, etc. The key tip- off is that they pay themselves first ... out of money your money earned.

    ------------------------------
    BARRY JOHNSON
    ------------------------------



  • 11.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-09-2025 12:17
    One more data point on Dr Bob's thread about AI bot accuracy.

    In July I subscribed to the AAII Investor Sentiment-Market Sentiment offering.

    It provides weekly updates to 12 indicators generated by AI bots that "analyze" the data that generate the 12 indicators.

    The outputs are very data rich with
    • current data for the specific indicators and related indicators that feed the key overall indicator  
    • ST, MT and LT data trends
    • comparative data
    • percentile rankings
    • average and outlier statistics 
    • historical probabilities from prior trends from 1986, and
    • connections, divergences between indicators.

    After little practice, I have reduced the weekly reports down to 1 page for each indicator, but even then, it takes a few hours and is still a headful that challenges short-term working memory.

    Wayne Thorp has written several articles on how LLM models CAN BE used to identify nuances, emotions, and changes in perspective communicated through the syntax, tenses, parallelisms, changes, and omission across industry, company-specific, and financial accounting, phrases, adjectives, and sentences - over time.

    I have a higher degree of confidence in data generated by AI bots that LLMs that parse text, but even here I have high level of skepticism about the potential to "guide" readers to speicific conclusions .. For "fun" and profit (theirs).

    Regards 





  • 12.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-09-2025 16:33

    Barry, I sure hope you're not paying AAII for that silly "sentiment" snake oil. Assuming one can correctly gauge current and past market sentiment (which is easy to do based on nothing more than market prices), how in the world is one to measure what really matters, which is future market sentiment? Surely you don't buy the notion that anyone can predict what the general public's (or the general investing public's) sentiment will be next week, next month, or next year! To even try to do that smacks of market timing (if not soothsaying), which any sensible investor knows is futile.

    The primary predictor I rely upon is the simple notion that the U.S. stock market will continue to go up over the long haul, provided our country remains the bastion of economic freedom and general ordered liberty. Beyond that, I rely on judgment and experience to tell me which companies are likely to exhibit long-term growth and profitability. Fortunately, I can be wrong (even most of the time) and still make enormous gains because of those few times I'm right. (I'm proud to say that in actuality, whether through luck or by craft, I've been right more than a few times.)

    For those who rely on sentiment tealeaves, I wonder how many times they can be wrong and continue to survive (not to mention prosper) financially. Where is the EVIDENCE that AAII's sentiment program makes anyone a better investor?

    (Sorry for my rant. My humble opinion of most of AAII's profit centers is pretty dismal.)



    ------------------------------
    Rob Adams
    ------------------------------



  • 13.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-09-2025 18:14

    Dr Bob

    As usual I always find your tips and guidance helpful.
    #1 I had a few bucks left over from my paper route delivering the Grit. I did take a deal. I'm sure they put it on the BB as the joke of the day. Can you take it out of my allowance or let me wash the car a few times?
    #2 I want to find out HOW the program works. I have read everything each week for 4 weeks to date (8/8). I find the combination of12 indicators interesting, but I haven't figured out how they ALL work together. Some I knew (AAII sentiment survey, SPX movements, yield curve, VIX, CAPE. etc.). Some new (TRIN, Fed Model, etc).
    #3 I am pretty sure the goal is NOT to "time the market," but to track indicator TRENDS and historic probabilities of what happened BEFORE and AFTER the current level of each trend so you get a sense where the market MIGHT be going and why.
    #4 I am a data junkie. These entertain me. I am learning a lot but, as Bob Seager sang, I am "working on mysteries without any clues."
    #5 The basic "link:" in this loosely coupled system is that "investor sentiments" are converted into "market sentiments" (through an undefined function through the interactions of these indicators adn then can be measured by other some of the market indicators. 
    Thank for the pep talk. I am not crazy. Juts curious.
    Regards,





  • 14.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-09-2025 22:14

    Okay, Barry, I'll let you slide this time. 😉 After all, Thomas Edison had to try thousands of wrong things before he found those that worked. 



    ------------------------------
    Rob Adams
    ------------------------------



  • 15.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-29-2025 12:14
    Edited by BARRY JOHNSON 08-29-2025 12:16

    Robert, 

    I posted this comment here because I could not post a comment to the article -- "Ideas From AAII and BetterInvesting for Finding and Analyzing Quality Stocks," pre-dated Sep 1, 2029 -- at aaii.com.

    I thought others would want to comment on the new "partnership" between AAII nd BetterInvesting.  

    =========================================================================

    John and Doug, thanks for the overview of what has been going on behind the curtain in Oz.

    #1 Screening OUT (skillful reduction and patience are also the keys to making a good roux) is THE key process shared by AAII (guru and factor screens) and BI (SSG template) methodologies.

    #2 So, focusing on what was NOT mentioned in this article may be the key to understanding what this Shakespearean "partnering" portends for US/YOU/ME. We know we can expect the fruits of the 'partnership' will be somewhere between (A) Benedick and Beatrice from 'Much Ado About Nothing' (a relationship based on mutual respect, John and Charles penchant for wit = very good for us, but conspiring to kill the other's enemies is also a shared trait) and (Z) Petruchio and Katherina from 'The Taming of the Shrew' (kidnapping, starvation and psychological manipulation through gaslighting = a modern-day 'Psycho').

    #3 I leave it to you to speculate on your own as to which parts curated herein will be screened in/out. [Shakespearean couples come to mind. I guess we will have to wait for the denouement in Act 4 or Act 5.] Based on the recommended due diligence process mentioned in the article, i.e.,  "when opportunities look promising, invest time in deep-dive research reading … and learning businesses as owners would, " my observations are.

    #4 Why this marriage between two companies with very similar philosophies? What's the "value add here? BI's SSG form seems to organize the outcomes of the AAII screens merely. 

    #5 "Value investing" has been around since Graham and Dodd in 1934. Both the BI (since 1950) and AAII (since 1977) have common "value investing" strategies.

    #6 So why now? Could it be as simple as agreeing to recruit each other's membership base openly? The article hints that AAIIers will need to join BI to gain access to the BI SSG methodology. [I paid BI $99 (50% off a 1-year subscription) for a free 90-day trial about a month ago.]

    #7 This article minimizes (i.e., completely omits any reference to) how the local BI club meetings contribute to the analysis process.

    #8 AAII's "systematic stock screening process" is designed to be an "individual investor" approach, while the BI club process is based on monthly meetings (which have separate fees and require you to participate in managing the local club portfolio.

    #9 My due diligence included viewing the 07/24/25 "AAII + BI: Smarter Sentiment Strategies" webinar, downloading and reading every document at the BI site and at the local club site, and reading about 20 articles in the very slick BI monthly magazine.

    #10 The big downsides here are

    (a) I don't do clubs because that's where FAs "fish off the company pier; "

    (b) reading all the BI documents; and, 

    (c) – the BIG ONE – reconciling the time and effort it will take to see an expected return that compensates me for learning "a second language."

    NET: I see how the AAII "Sentiment Investing" adds to my investing skillset, but I don't see what "new skills I will learn from the BI process, forms,  and clubs … yet. I still have 60 days to go on my trial membership before I will have to renew at @$199.

    Regards,



    ------------------------------
    BARRY JOHNSON
    ------------------------------



  • 16.  RE: Can AI Really Help Value Stocks? Let's Talk Large Language Models in Investing

    Posted 08-29-2025 16:00

    Barry, I'm a member of AAII for the same reason I subscribe to Kiplinger's Personal Finance magazine (and have, over the years, subscribed to a number of other publications). Through each, I occasionally find some useful information I can use. It happens less and less, but it nevertheless still happens. 

    Aside from the occasional nugget, these message boards have helped me a lot. Writing helps reinforce and clarify my own ideas on investing. It also helps to read the nutty ideas others have (including some of the "professionals" at AAII) so I can avoid getting caught up in them. 

    From what I can see on the Better Investing website, they're more equities focused than AAII, which is a good thing. But I'm not going to pay their higher membership fee for the hope of more occasional nuggets. Also, I don't need to become a member of yet another organization that merely uses my membership status as a means to try to upsell me. 



    ------------------------------
    Rob Adams
    ------------------------------