How to Use ChatGPT4o as Your Stock Analyst ($NVDA, $GOOG)

[@tasadem20]

Great way of using technology. Thank you for sharing.

[@finxter]

Thanks, appreciate it!

[@yapyh2872]

Hi, i really like your information. Thank you. I'm totally new to stock assessment but i did try it out and find out from gpt 4o that there are also a few type of DCF analysis such as a Risk-adjusted Cash Flows, Unlevered Free Cash Flow, Detailed Multi-stage DCF Model and etc. Which would you recommend to use and which is more accurate?

[@finxter]

I like simple unlevered FCF model. It's not rocket science and people overcomplicate it. I'd rather be roughly right than precisely wrong. ;)

[@surelywoo]

Great presentation. I didn't know it could do this.

[@finxter]

Thanks man 💪

[@danesaberlinesa]

Great video mate... ive been thinking for a while about how to use the new GPT versions to help me and assist with stock analysis. Its great to help you spot some insights about the stock that we, as amateur investors, otherwise might overlook... and its a way of keep learning and sharpening our stock analysis skills. Thanks for sharing

[@finxter]

Thanks, great PoV. Agree 100%!

[@SantezFlatline]

Thanks for the video. I used the same version of the gpt chat 4o, the same request and the same screen and got different output results. I can’t understand why, you have more information about company in your answer, I have less text in my answer.

[@finxter]

Yeah, they always update the model (even within a "macro" version such as 4o). Also it's non-deterministic output...

[@rafref]

Thanks for sharing, have a nice dayy

[@finxter]

You too :)

[@Terencethys]

is the current GPT3.5 VERSION able to assist in stock analyst and current buy levels based on your illustration?

[@finxter]

No. ChatGPT-4o is a 10x improvement because of its integration of deep RAG functionality in my opinion.

[@Terencethys]

@@finxter does it have real time data access to investment resources to assist in calculating fundamental strong companies? how should I fully utilise?

[@Terencethys]

does Chat GPT-4o has its own biases when doing our information research and to do our due diligence?

[@techwithlaki8917]

Thanks for sharing

[@DividendDynamicsYT]

I made a video where I compare the AI with a manual analysis. Great video!

[@finxter]

You could track AI and your prediction accuracy over time. Could be interesting.

[@pavanghai]

Which tool are you using to capture full screen of yahoo site?

[@finxter]

Haha, just open it with Firefox, right-click to open the context menu, select "Take Screenshot". No need to install any external screenshot tool.

[@MyBrianBailey]

you can get snipping tool for free on PC

[@maxlewisconsultantspteltd7187]

@@MyBrianBailey After snipping, how do we paste onto the chat gpt4 ? can someone help?

[@egorge00]

Great video, thanks ! Any means to automate this through a number of companies ? Such as: asking to make the same analysis with all the nasdaq 100 companies and get a recap tab ?

[@finxter]

Definitely! Simply use the OpenAI API for raw intelligence, and feed it the financial statements via an API (e.g., Yahoo Finance API). Great idea. I sense a million-dollar business idea!

[@rc-y]

@@finxter😂

[@geno755]

Question: how current is the data when querying chatgpt? Isnt there a delay?

[@finxter]

They use finance web API (seemingly) so pretty current in most cases. But it's always good practice to cross-check, especially if you base some investment decisions on it.

[@Barabus-yx2cn]

It's outdated sh!t.

[@ASHAH925]

How do you use chat gpt for options trading?

[@finxter]

Not a big fan of options "trading". Buying long-term Leaps maybe - but trading sounds like a loser's game. Better buy stock (with theoretically infinite holding period and no time decay).

[@sedul2006]

Is there a way to encapsulate the prompt/automate the prompt? Train a chatbot to run the same analysis on the same input?

[@finxter]

Yes, sure. One could use fine-tuning: blog.finxter.com/openai-fine-tuning/

[@larsthonissen8295]

Nice Video! I do think you can also just tell GPT to check the URL. You don't have to feed it a screenshot.

[@finxter]

Not anymore since the recent copyright lawsuits. At least in Germany - feel free to confirm if it can read Yahoo stock information in your country.

[@Soulastro12]

can you do AMC?

[@finxter]

Maybe in the future. Thanks for your interest!

[@ahmadhuseynli2073]

Can we trust it for our own investments ?

[@finxter]

Don't trust. Verify. ;)

[@loureiro7342]

does microsoft ai (copilot) does the same?

[@finxter]

They're probably using ChatGPT internally so yes. :)

[@bonanzatime]

What do you mean by, '15% discount rate'? What are you talking about?🙁

[@finxter]

"The discount rate is the interest rate used to calculate the present value of future cash flows from a project or investment." I asked ChatGPT to explain like I'm five: Imagine you have a lemonade stand, and you want to know how much it's worth today based on how much money it will make in the future. The "discount rate of 15%" is like saying future money is worth 15% less each year. We use this number to figure out how much all the money you'll make in the future is worth right now. It's like using a special rule to say, "I’d rather have money now than later, so future money isn't worth as much." Because I like money now more than money in the future (=future cash flows), I reduce the value of future cash flows by 15% each year. This is a very conservative approach.

[@bonanzatime]

@@finxter Ok,🤔 it sounds like you/it is describing inflation without using the word inflation. Is that correct?

[@tmiris3006]

@@bonanzatimehes not describing inflation. ”Disccount rate” is a term to discribe the value of future earnings. Think about it like this: if i told you i will give you 1 thousand dollars ten years from now, what is the value of that right now? Well it isnt the full one thousand, because its not as valuable to you if you dont have it right now. So its worth ”some percentage” of that one thousand. That percentage is the discount rate.

[@lky88lim38]

15% discount…. Is this realistic? I wonder which stock is not considered overpriced by this yardstick? 😅

[@serknight029]

​@@lky88lim38it's not. With current interest rates and historical equity risk premium a normal wacc's range is between 7 and 9%

[@kaytoonen6266]

Lol it just did the annually estimated growth in the last month around 40% chat gpt was not bullish enough haha

[@finxter]

Haha, yeah. Hindsight is really useful for financial analysis. Even we have an advantage against AI if we have the benefit of hindsight. ;)

[@icholakov]

The fundamental analysis here is straightforward but not very practical. It has been performed for decades without the help of AI. In fact you are not using AI for any insights but merely for a more natural language interaction with fundamental data. You are valuing a stock at a point in time which in by itself cannot be used for investment decisions. For example, if you conclude a stock is significantly undervalued, you cannot invest in it because it can stay undervalued for the next 10 years. Similarly when a stock is overvalued it does now mean its stock price will slow down.

[@finxter]

Yeah sure, it's just a tool that simplifies valuation work. The future will bring down the barriers of company valuation so retail investors will keep pushing into the industry - many will outperform so-called "professionals".

[@dimashlapakovsky3779]

Good use of AI BUT you inputs are wrong that’s why your numbers are off. You can’t just pull those figures off Yahoo finance and call it a day. DCF should be based on unlevered after tax cash flows.

[@finxter]

Thoughtful comment. Just a quick note - there are many ways of running DCF. I agree that numbers should be chosen carefully but it's generally a good appraoch of running various types of analyses, even more heuristic "napkin math" types. If it was all about precise accounting, accountants and math PhDs would be the richest people on the planet. Everybody needs to find their individual strengths in investing, all input numbers are wrong (even yours in your DCF analysis), and it's not the best approach to use "unlevered after tax cash flows" for every company in every situation. I'd argue that using net income for high-growth companies is often more accurate than FCFs due to the investments in growth infrastructure (e.g., GPUs).

[@niknik1604]

Can someone take this prediction serious ?

[@finxter]

No, you shouldn't blindly trust any output. But you can take it more serious if you put in your own numbers and assumptions. This is just a tool, not a crystal ball.

[@diebestenaktien2323]

If Nvidia grew 42% pa, the profit in 10 years would be 1 Trillion……how probable is this? Zero!

[@finxter]

Why shouldn't this be possible? People 10y ago believed Apple or Microsoft's profits and marketcap is impossible. There's no physical law that prevents companies from earning $1,000 billion in a given year. In fact, I believe it's highly likely given the debasement of USD and the winner-takes all economics in today's world.

[@Huelosoneroen]

Jo, die beiden Stonks haben halt in den letzten 10 Jahren 28% bzw. 25% p.a. hingelegt

[@aosaller]

​@finxter I'll tell you why. Because to have profits that high, even with current high margins it would need to have revenues north of 2 trillion. And the chip business is highly cyclical. Jensen himself said there's a total global AI Capex budget of 1T, and that is spanning multiple years, of which nvda could maybe get 500bn. If you look at consensus estimates, you are getting 140-150bn estimates annually for the next 5 years, so nowhere near your assumed amount. Plugging a bunch of figures into chatgpt and having it spit out a number that you like doesn't mean one's analysis has any merit. Thorough fundamental research is what's key, and ai is just a tool.

[@michaelherrlich1536]

The calculation of the implicit growth rate is not so useful as you add your 15 percent in the calculation. As the implicit growth rate is 26 percent and you want it to compound by 15 percent, the earnings growth must be pretty low compared to it's currently high price.

[@finxter]

Thanks. Would you be able to clarify this please? Not sure I get what you're saying and before I reply I want to make sure I do. We need to differentiate the growth rate of the business itself and the return of investment when investing in the stock. That's why we first estimate the growth of the business to get its intrinsic value. By adjusting the discount rate we can get our personal ROI from investing in the stock when our business growth assumptions are correct.

[@philippeturco4670]

LUCY IS THE NEW MEME STOCK🎉🎉🎉🎉🎉🎉

[@finxter]

Don't know this. Don't buy meme stocks.

[@zRoGuee]

Seems you were correct lol how did you know it would be? Please share your sources :)

[@michaelherrlich1536]

Warren Buffet is not doing any DCF. Read the books. Buffet takes some main variables, thinks about moat, management and the future, and then buys great companies to hold them. Great moment to buy when they're under distress. Your video is nice to show what chat gbt can do, but don't bring Warren Buffet here as a reference

[@finxter]

Strong disagree. Warren Buffett has said in multiple shareholder meetings that DCF is the only way to value a business. He's using "owner's earnings" instead of "free cash flow" or "net income". He says that he's doing a mini DCF in his head and buys companies only if they're at a significant discount to "intrinsic value". He defines intrinsic value as all the cash a business makes from here to judgement day, discounted into the present, which is exactly the definition of DCF.

[@darrenting]

Isn’t the 2024 growth rate is well over 100%? 2025 will be somewhere around 60%

[@finxter]

Can you elaborate? NVidia revenue growth rate over 100%? I doubt it. I'd say roughly 50% if you forced me to guess...

[@jonnyd95]

Schönes English 👍

[@finxter]

Haha, do I sense sarcasm?

[@averagescav1947]

no front you are way too bullish look at the following interest rate hikes in the past and cisco 2000 and today with nvidia 🙂 im going to spare you time it is the same as cisco in 2000

[@finxter]

So? Not sure we should reason by analogy here using a single data point as analogous scenario. You can always overlay technical charts and conclude anything from it. There is little information gain approaching a qualitative analysis from a TA PoV.

[@ppp123ppp]

Overvalued as f#uck. Just like Tesla. Lets see both stock performances in 10 years.

[@ppp123ppp]

I think both of them will be flat or lower in 10 years

[@finxter]

Thanks for the perspective. I don't think so at all. No way flat or down in 10y. But that's what makes it interesting!

[@ppp123ppp]

@@finxter That´s the point. Everybody think there´s no way that could happen. The market punishes consensus

[@dumaoctavian8523]

Be carefull, you are way too bullish

[@finxter]

You think? I'm not sure - it is a revolution. Scaling intelligent labor is not a toy and it won't go away. It's like the industrial revolution on steroids. I think most people are just too stuck in their ways that they underestimate the disruptive power that comes with AI.

[@bar590]

@@finxter agreed!

[@Sujit_Kulkarni]

Bullshit

[@timmark4190]

Why.

[@Sujit_Kulkarni]

You don't need gpt to do dcf for you. There are tools already that can do it easily within mins and investing isn't that simple else everyone who is Powerful with money and brains would have done it without you and I knowing about it

[@finxter]

This is about the tooling for investment analysis. Go ahead and mention some tools that are better for DCA than an intelligent, data-aware, financial analyst that you can iterate with to include or revise your own expectation for a certain company. It is my view that this foundational model is the best tool out there to help you with financial analysis. Certainly better than asking your "trusted banker" who's less intelligent, less skilled, and less objective than ChatGPT-4o.

[@verngall]

@@Sujit_Kulkarni If you don't like the content just leave. no need to hate on the author ->. please. Or create your own videos.

[@Mark-bb4ks]

Looks like it wasn’t bullshit. Nvidia sky rocketed

[@Barabus-yx2cn]

Use Perplexity. Far superior to this garbage.

[@finxter]

Not really. Perplexity is great but it uses inferior models and RAG compared to ChatGPT-4o. Just check out any model comparison benchmark online. Not sure why you call ChatGPT-4o garbage?