How I Use AI To Trade The Markets! | Kris Bullock and Bijan Maleki

0:05Bijan Maleki:Happy Wednesday, everyone. Welcome back to another edition of Trading the Markets with AI. And we've got a great show for you. We've got Real Vision member Michael who will be joining us in the second half of the show to talk about how he's using AI with his workflow and his personal life. You guys know the theme on this show. We bring on real people to talk about real use cases that they're doing with AI. But before we get into all of that, you should also know that the arena is live and kicking. We have so many trades coming in just left and right. People are already up like over 50, 60 percent.

0:49Bijan Maleki:Some people have been trading options as well. So you can join Real Vision for$20, join RV Connect for$20. You get access to everything that RV Connect has, which of course is Chris Bullock's portfolio. You get Joe Bland's portfolio. You get our Monday show. You get our monthly AMA with OSF and Mando. And of course, you get the show with Joe Bland as well, where he talks markets. So join RV Connect for$20. And you can also post your trade idea, one trade idea per entry. So let the market decide who's right. Want to see some great trades coming in. And of course, we've got some news in the AI space that we're going to dive into.

1:36Bijan Maleki:If you've been considering using AI as a financial advisor, we're going to talk a little bit about that. But looks like Anthropic and OpenAI have some competition as well with a new AI company entering the fold. And they just touted their new release as well. And of course, we also have AI and privacy coins that were kind of leading that rally that we saw as Bitcoin was retesting 75K. So, you know, we'll probably get into that as well. So Chris, happy Wednesday, my friend. How are you? Doing well, doing well. I'm looking forward to this, definitely. I wanted to touch on this first AI article that I read because it speaks directly to what we talked about last week with prompting and prompt engineering and things like that.

2:28Bijan Maleki:And how, especially if you're using it for research, for financial research, you can't just go on and use ChatGPT like a Google, you know, like you're asking Google something. You need to actually really refine what it is that you're asking for and maximize the constraints so that you're not getting hallucinations, so that you're not getting just factually incorrect information and basing major financial decisions off of. So this article came up this morning, and basically it's saying that 66 % of people out there, 66 % of Americans have used generative AI for financial advice, and younger generations in particular are more likely to use them with 82 % of Gen Z and 82 % of millennials using AI for financial guidance.

3:16Bijan Maleki:So again, it comes down to, you really need, go back and watch last week's episode about prompt engineering because you have to actually make sure you know what you're doing here. Otherwise, you're just going to be flushing your money down the toilet, basically. You're going to get poor information. You're going to ask the wrong questions. It's going to give you factually incorrect information and it's going to end badly. So I think there's definitely a place for using AI for financial advice. Believe me, I personally do it myself. But again, you need to know how to do it correctly. Otherwise, it's bad.

3:52Bijan Maleki:And it's a good thing that they do know how to do that because we just talked about, like you were saying, we just talked about that last week with the prompting, right? And making sure you're getting the right data with the APIs that you want to use so you can guarantee that it's not just making things up. Because sometimes what I've noticed AI does, if it can't get the data, it makes it up with believable data, which you got to be careful. Yeah, it can sound very convincing. It can sound very much like it's speaking the truth. And it's not sometimes. It's just plain not. And don't just like ask it one question.

4:29Bijan Maleki:I mean, ask it a series of questions. Actually ask it to fact check itself. Ask it to reference specific data. You know, ask it to compare things and provide context. Like I said, act as the role of a hedge fund manager or a research analyst or things like that. Provide the context, you know, so that, again, so that you can really refine the output as much as humanly possible because otherwise, yeah, not good. Not good. And you know what I found, Chris, that actually helps with that at the end of the prompt, right? Obviously, you want to get the prompt right. You gave us some good tips last time.

5:04Bijan Maleki:You prompted to be a certain profession, be a financial advisor, and then dive into what you're wanting to get out of it. But I found that at the end of that prompt, when you can say, after you give me the output, I also want to know the sources that you use to gather this information. And then it'll list the sources that it used to be. You can also verify it as well and also make sure it's actually using good, legitimate sources. So that's something else you can consider when prompting. Yeah, absolutely. A hundred percent. Just like I said, go back and watch last week's episode. We spent 10 or 15 minutes talking about one way you can engineer a prompt, particularly like if you find a YouTube video or an article that mentions a lot of things and you want to look at this and go, how can I distill this into actionable information that I can express in a trade or something like that?

5:57Bijan Maleki:And this series of prompts will break it down and just give you step-by-step as a way you can do that and give you some other resources for finding additional types of prompts that you can use for various research techniques and things like that. And if you guys go back to our last Wednesday's video in the comments section, Chris actually pasted his example prompts in the comments section. So you can use that as an example to help you frame your prompts as well. Right. Exactly. Exactly. So next article. This one was really interesting too. And I can totally see this happening. Basically, 63 % of workers say AI will make the workplace feel less human.

6:39Bijan Maleki:And that's becoming sort of a greater fear than the humans losing their jobs, necessarily, is that rather it'll make it seem like a robotic, heartless environment where less critical thinking involved, less humanity involved. less just, you know, human experience. And I think that that's important and I think that's relevant too. And I think what we're seeing right now is of course, because this revolution is blowing up, we're seeing the pendulum swing way, way out into everybody trying to use AI for everything. And I think eventually that's gonna swing back and we're already starting to see, you know, like analog movements.

7:21Bijan Maleki:We're starting to see people like just write off the whole thing entirely and embrace old analog technologies and more simplified ways of living. And, you know, it's causing people to sort of detach from a digital life to a larger degree. And I think that that will happen. We're seeing sort of the opposites come from this. But I think that this is an important insight here. And I don't disagree. I think that this will happen and it'll probably get worse before it gets better. But like I said, eventually things will sort of swing back in the other way and we'll find an equilibrium at some point where, yes, there is a use for AI and it will be apparent, but it's not going to be everything.

8:02Bijan Maleki:And it's going to take a little while to unfold, I think, too, which is unfortunate, but that's what happens when we have these major technological revolutions. It blows up and then it comes back down and then everything sort of normalizes and we find this equilibrium going forward. But I think we're going through that right now. yeah i mean i guess if you're an introvert that might sound uh attractive to you less less people in the workspace but you know it's not ultimately we are we are creatures that need company right humanity needs company so uh definitely something to watch and and also lends to you know i i comment this a lot of what ral has said in the past that you know as as our society it gets increasingly technological, these human experiences are going to become increasingly more coveted and sought after.

8:50Bijan Maleki:So that's also something to keep in mind, guys. Yeah. And I think it's going to express in a lot of different fields too. I think we'll see it in creative fields. We'll see AI start to enter into music and art. We're already seeing it now with a lot of AI music coming out. But I think that that's going to help spur more analog music. I think we'll see a shift back to people actually buying guitars, you know, and starting bands and actually trying to recreate that human experience. And, you know, that'll be competing directly with AI because, you know, we know AI music isn't going to, I mean, there's a small, very, very small contingent of people that are going to embrace that, you know, but ultimately people are going to be like, I don't want to hear this.

9:35Bijan Maleki:This this isn't real, this isn't creative, this isn't a creative expression and that's why I enjoy something like this or art or paintings or what have you. And so I think it will, like I said, spur a movement back into more the human side of those creative endeavors as well. Yeah, and I mean, and I think too, like we said, those human experiences are gonna be more sought after, but as AI content becomes increased as well, human content, I think is also going to be more sought after. And I could see people paying a premium for like a human written book, as opposed to like, you know, something that AI wrote, because they actually want that connection to another human being.

10:18Bijan Maleki:And I think one thing where this shows is emotion, right? People can always tell when you're not being your authentic self, right? They, you're like, they can tell if you're faking it, or you're not being who you are. And I feel like that's the same way with AI, especially when it comes to emotion, if you've experienced, like, I don't know, you know, the death of a loved one or like a really like important milestone in your life that has a different like feeling when you write it out than AI who's never actually like known what that feels like. So I think that's, that's something else that can be very, you know, powerful in our future as well as human written content and created content, obviously.

10:56Bijan Maleki:Yeah, yeah. Honestly, I kind of can't wait for that to happen. I miss, you know, human created music and we haven't had a lot of innovation. I feel like in the last few years, it's been very homogenized, you know? And so I'm, I'm looking forward to some, some innovation coming back to that arena. Okay. Finally, moving on. This was an interesting article too. So Mistral bets on build your own AI. And you had touched on this at the top of the show. Basically this company, Mistral AI has developed an open source AI for companies that want to keep their information private, in-house, but still want to benefit from AI technology that can't sort of adopt public models, public cloud-based models or anything like that, that don't want their information out there.

11:40Bijan Maleki:They can utilize this technology to basically train a LLM on their internal workflows and their internal data. And you can have it highly refined and specified to whatever your industry is. So it can be very effective at just that one thing, which is great. And also it's perfect for like non-English speaking. You know, it can be, again, refined to be custom tailored to very niche use cases where it excels as opposed to just a broad based LLM that's kind of good at everything but not great at everything. You know, this can be like really great at one thing and very specific to your business. And so I think that there's definitely a market for this.

12:19Bijan Maleki:I think this is going to do well. And yeah, not surprised at all to see this. Yeah, I mean, I feel like it was obvious in hindsight that something like this would be coming when we're talking about people being laid off due to AI and companies embracing AI and using that more increasingly in their workflows. like obviously an agent you know or an ai you know llm protocol specifically catered to the enterprise level was bound to happen and it was just such an obvious thing because like you said chris these data is very important to to big companies right and they don't just want to put that data out there a lot of it is confidential it's proprietary they don't just want that out on the internet and unfortunately if you're using chat gpt or or claude right you have to you should know that any conversation you're having is like a public, could be like a public conversation, essentially.

13:13Bijan Maleki:And this kind of addresses that. So I think this is good for competition, and it's just good for this agentic space that we're building out. So definitely want to see how enterprises implement this and see where this goes. Yeah, yeah, for sure, for sure. Okay, moving along, I want to, so for this episode, I'm going to demonstrate what I've been doing with AI over the last week. And this is, I basically had it build me a daily executive intelligence briefing. And I'm going to kind of walk through what I'm having it do here. And some of the prompting that I went through to get to where this is. And I've got it way zoomed in.

13:54Bijan Maleki:It's normally very tiny. I actually look at it on my phone, but for illustration purposes, I've got this maxed out. So what I did is I follow a bunch of YouTube channels regularly, in fact, 24 of them to be exact that I follow on a regular basis. And I wanted a system that would every morning go out and scan all of these channels and basically find news information that is relevant to my portfolio. So I was like, I need to give you my portfolio. Here's all the symbols I'm watching. Here's all the symbols I'm holding. And then also here's the list of YouTube channels that I want to follow. Go out every morning, find all the news from all those channels and tell me which of my assets are impacted by any of this news, whether it's bullish or bearish.

14:41Bijan Maleki:And if so, why? That's kind of what I wanted to know. So that way I don't have to go and scan all this stuff. So I didn't really know exactly how this was going to look. So I actually, it's funny, a lot of people have been using OpenClaw. A lot of people have been using Claude code. I've fell into a routine where I actually have the most luck with just using the basic Claude chat. Like I still use the desktop app, but I just go into Claude chat and I start having a conversation with it. It's really as simple as that. I just ask it, you know, do you have the ability to perform agentic work? I'm interested in creating a dashboard that goes out and collects this information.

15:18Bijan Maleki:I want to install it on my VPS so that I can access it from any device. Can you do this? You know, that's how I started. And it said it can do that, but the only problem is it can't do the scheduling. But since I'm using a VPS, it can write a Python script basically that does a cron job that can automate the scheduling for me that way. So even though it's not using like cron jobs within OpenClaw, you can still do scheduled events outside of OpenClaw, you know, and have sort of agentic functionality just with Python code and scripting on a VPS or whatever. So I was like, great, perfect. So I kind of went back and forth and had some conversations.

15:53Bijan Maleki:And then of course it suggested we can do YouTube, but you know, also there's some great APIs out there. There's a bunch of RSS feeds. Do you want me to collect news from other sources as well? And I thought, yeah, okay, sure. Why not? You know, like, I mean, that's ultimately what I'm looking for is the most impact on my portfolio. And so the more information, the better. And so, yeah, so basically I just went back and forth. It collected some APIs. We had a few glitches with getting the YouTube API to work. This was interesting too, because one thing aside real quick, I had been using Gemini for this task initially because Gemini is a Google company.

16:30Bijan Maleki:It's integrated with YouTube already. It can go in and pull the transcripts with minimal fuss. But it didn't even occur to me that YouTube also has its own API that you can call and pull scripts from that way. So I was like, I wanted to try this with Claude and see what Claude could do with this. And then finally, I wanted to collect all this information, but I don't want it to just collect it and present it to me. I want it to actually process the information and summarize it and give me kind of, again, a briefing, a daily report. And so it's interesting. It actually suggested I'll use the sonnet model to kind of collect everything and do the general quick summaries of each video.

17:10Bijan Maleki:And then I'll use Opus to actually prepare you like a final judgment report. I'll use the deeper thinking model to prepare a final sort of here's everything you need to know takeaway kind of thing. And so, yeah, so this is it. And basically what it does, it's broken up into a few different sections. Initially, it just gives me kind of a TLDR right at the top. Like here's the main issues. We're going on in macro, tech, liquidity, geopolitics, et cetera, crypto. So those are kind of the big TLDR takeaways. And then the next section down is actual summaries of each news source, whether it's a video or an article from RSS or whatever.

17:52Bijan Maleki:And it gives me like a summary. And then I wanted it also to focus specifically on the transition in AI from, you can see there's a lot, which I'll get to in a sec. I wanted it to focus on the transition between broader cloud-based LLMs that we're all using now into the next sort of wave of AI being sort of edge devices and endpoints and local inference happening on our phones and on our computers and on our home appliances or whatever, the stuff around the house. because eventually AI is going to move to that where we're not going to use these cloud AI systems as much. And a lot of the actual AI compute is going to take place at our edge devices.

18:36Bijan Maleki:And so I want to benefit from that transition. And from an investment standpoint, I want to be exposed to the, because we've already kind of missed the run-up in semiconductors and chips and stuff like that. And so I want to move, I want to be ahead of the game and catch the next evolution of AI into edge devices. And so that's why I have a special particular focus on that. And then finally, it goes out and it takes all of the inputs from all of the different talking heads, basically. And it establishes like a breadth consensus. Like, does everybody think that it's bearish right now? Or do only half of the people think it's bearish right now?

19:16Bijan Maleki:Or is everybody bullish right now? You know, what's kind of the across the board? What is everybody thinking? And kind of give me a score of that. And so what it does is over on this side, finally, at the end, is right now the source consensus, the macro breadth consensus is actually pretty 50-50. There's no dominant expression amongst today's articles, which is interesting. Yesterday it was, I think, 78 % or something like that. But right now it's 50-50. There's a lot of people that think things look good. There's a lot of people that think things look bad. But it's important for me to know that so that I can kind of factor that into my evaluation of this.

19:53Bijan Maleki:And then finally, over on the right side here, it has my watch list and my portfolios. And so each asset that was mentioned, it tells me, you know, how many times it was mentioned, what the output was, was it bullish, was it bearish, what specifically is going on, and why I think it should be important. And so like in this case, we'll just look real quickly. Micron, bullish. They have their earnings call today. You know, big announcement with NVIDIA and, you know, et cetera. And anyway, you get the idea. So I can go through this and look and see what looks good, what looks bad. Interesting, gold has been popping up a lot lately.

20:33Bijan Maleki:I know the chart doesn't look good. I know gold has been kind of faltering lately. I know Bitcoin's gained a fair bit of ground on gold lately. but everybody in this list for the last several days has been really bullish on gold ultimately prevailing under the current geopolitical circumstances that we're dealing with. And it's only a matter of time before the trade comes back around to that. And so that's something that I've made note of. And so, yeah, this is something that I look at and you can see it's a lot of information. And at first I was like, whoa, this is too much. Initially it was just the YouTube and I was like, this is good.

21:09Bijan Maleki:I can whip through this in three to five minutes. But after I read it a couple of times, I was like, you know, this is actually better because it still only takes me maybe 10 or 15 minutes to read through. And it's a lot more information. It's a lot more broad scope and it factors in a lot more insight, you know? And so ultimately, I want the best signal out of all of that. And I'm fine to maybe spend an extra five to 10 minutes reading this report. I mean, in the end, it's still only five to 10 minutes out of my day, as opposed to taking what two to three hours to process all of this, you know, on my own, I would much rather spend 15 minutes of solid just like, okay, I'm good.

21:52Bijan Maleki:I know what's going on everywhere. Now I know who's saying what I know what looks good. I know what looks bad. And I can I can, you know, proceed accordingly, for you know, for the rest of my day. So I think that this is kind of the sweet spot. It's not too big. It's not too small. It's kind of just right. And it's very, very in-depth. And I love it because it's specific to my personal investments, you know, and that's what matters the most. And so, yeah, this is what I've been doing. One of the things that stands out to me, like almost right off the bat, and that I absolutely love about this is you basically, or I did, build a personalized fear and greed index, just strictly based on your portfolio, right?

22:36Bijan Maleki:Instead of like, when you look at the fear and greed index, you just see it as just the market as a whole. And yeah, it tells you some stuff, but like this really is like dives into individually, right? Like Micron, that's greed, right? It's nine out of 10, right? And then, oh, that's so cool. So you have a great indicator stock by stock. So it's not just your portfolio as a whole, but then I like, cause you do have that where it shows that 42%, But like, and then it also gives you individual per asset that you hold. That is so cool, Chris. Now question for you. Could you mention, and I have a feeling a lot of people will be like pulling the YouTube API, right?

23:16Bijan Maleki:And you mentioned you had a little trouble like with the YouTube API. Like, and I found myself also having the same issue. How did you resolve that? Or like, where are you at with that part? Well, yeah, it's a little clunky. Basically, YouTube has like, I don't even know, 50 different APIs. And so it's like you go on there and you're just like, what do I even need? Which one of these is even relevant to what I'm actually doing here? And then even once you get that API, there's all kinds of checkboxes for specific types of information that it can collect from within that API data set. And I didn't necessarily know that either.

23:57Bijan Maleki:So I was having to ask Claude, what do I do here? Like, what is he even asking me for? Which of these boxes do I check? What's relevant? What's not? And then once I did that, because Claude had added so many other things that I'm collecting, it ended up like kind of hammering it. In the end, as I was developing this, you're only allowed like 10 ,000 whatever tokens per day or something to calls against this. and each time I would run this collection process, it would do like 2 ,400 tokens. So in the end, after four tries, it would max me out for the day and I couldn't keep testing on it and developing on it.

24:36Bijan Maleki:So I'd have to come back the next day. So like two or three days in a row, I maxed out the YouTube API just in sort of the development process before I could get it dialed in. Now it's fine because it only hits it once a day and now that it's working, it's good. But yeah, it just took a bit to kind of figure out what exactly you need and what information was relevant, what checkboxes are relevant. And of course, YouTube being YouTube on the back end, the developer console, it's not intuitive at all. You know, it's pretty clunky. there's a lot going on there you know um because it's not just youtube it's like google you know and so it's it's kind of everything that's sort of within the google sphere and yeah it's it's but again just just ask claude and it'll tell you you know that's all you got to do if you look at it you open it up you're like what the hell am i even looking at here just type you know hey what is this i don't know what to do where do i go what do i click you know and and it tells you of course So it's really that simple in the end.

25:32Bijan Maleki:But yeah, it was clunky for sure. Well, you know, a trial and error, right? Even though AI helps us get things done fast, still got to tinker with it, still got to do experiment and make sure you get that stuff right. Chris, this is such a cool little portfolio tracker you have here, Intelligence Hub. I'm going to, you've inspired me because right now I have it generating reports and it sends it to me as like a as like a page you know like a five six page report but i like this so much better so this weekend i'm going to be copying you chris i'm going to copy your vibe codes because this is so cool um guys if you have any questions about how chris built this you have any questions about like apis or anything because it can get a little confusing especially we're not used to using apis and like chris said youtube has like so many different apis there's the off uh key as well that you might need in certain cases.

26:33Bijan Maleki:Twitter has a few different APIs. They have the client secret key. Then they have the API key as well. There's a few different things going on there. So if you have any questions, obviously ask your LLM too, but we're here to help you as well as people that have gone through it ourselves. So keep the questions and comment. Yep. I just want to touch on something you said. I'm glad you're inspired. And that's really, I want to reiterate the whole point of this show that we're doing is to inspire our audience to use AI to build things. And last week it was the complete opposite. I was wildly inspired by what you had created and what you presented.

27:09Bijan Maleki:And I thought it was really cool. And I'm like, this is what I was hoping for when we created the show was just, let's show what we're doing. Let's show how we're building and let's inspire this audience, this community of ours to do this, to get out there and start building. And so I'm so glad that it resonated. And yeah, that makes me happy. Yeah. You know, I think as fears are, you know, with, with AI replacing, you know, jobs, those fears are, you know, certainly warranted. Right. But there's also the other end of that, that people that learn how to use AI and learn how to work with AI to improve their process, they're going to be like so much more valuable in their jobs than, than if they didn't.

27:49Bijan Maleki:And that's us here. That's everyone in this community. That's everyone who watches this show, that's what we're doing. We're all learning together on how to get better with AI so that we are just crucial instrumental parts to our jobs. So guys, job security lives here on this show. So anyway, Chris, thank you so much for showing us that. We're definitely going to check in with this. And I know people in the Discord are going to be asking you all about this. But let's head over to Michael, who's been waiting so patiently for us to show us how he's been working with AI. So, Michael, if can you just really briefly tell us like what you're going to show us?