How Real Traders Use AI to Trade
Okay, everyone, welcome back to another episode of Line Your Own Pockets. I'm personally pumped with this one because we're finally talking in-depth about AI and something that I'm getting super passionate about the more I play with it. And Dave's got a really good example of something that we talked about in the last episode that he's been building and playing with and how he's used AI to really help him do that. So, Dave, you're you're drinking the AI Kool Aid. You're you're half robot now.
Michael:How's it going for you?
Dave:Yeah. It's I I really used it extensively for this last project I did, and it was the first one where I really it was the first major software project I took on with ChatGPT in my side pocket. And, yeah, I was kinda blown away. So, yeah, wanted to I thought we should do an episode on it. So first set the stage
Michael:Well just first, so ChatGPT, that's what you used. Was there any particular reason for that one as opposed to the other one or did you find that they're all kind of working the same?
Dave:So I've used Claude quite a bit and I use ChatGPT. I found that probably the only reason that I chose ChatGPT, thinking back on it, is the free limitations that you get. Like, you get, you know, free access, but then blah, it cuts you off, it seems like, so quickly.
Michael:Very annoying. Wasn't.
Dave:So I just found myself using ChatGPT a little bit more. And I had paid for it when it first came out and then I'd canceled it. But then I paid for it again. I'm like, wow, okay. This is a perfect use case for this.
Dave:And yeah, that's really probably the only reason I chose to be honest.
Michael:Yeah, because I just wonder from, as someone who's not a dev, exclusively use ChatGPT and I have my own reasons, so I'm sure we'll get into. But I always just wonder how big of a difference is it? Because it's one thing that I commented on Twitter that people really responded to when that deep sea came out, And it had that deep research thing that ChatGPT didn't, but but now it does. I was talking about how I just instantly switched and how there's there's seemingly like no brand loyalty or any sort of moat between any of these guys that I sometimes I'm on Gemini, sometimes on ChatGPT, and you can just kind of pop around. Yeah, so I was just interested for for why, but that makes perfect sense.
Michael:I found the same thing with Claude. Gets very obstructionist very quickly. It's the poking to get paid happens way sooner than I find in the rest of them. Yeah.
Dave:Yeah. There was a specific example a few months ago where I had a very specific problem I was trying to get CheapGPT to solve for me. Was I needed to read this algorithm for a script I was writing. Mhmm. And ChatGPT could not get it no matter how much I was sort of leading it in the right direction.
Dave:It every time it got it wrong, I was like, Man, let me try it with Claude. Boom. Right away, the first answer it gave me was right. So that really piqued my interest on Claude at the time. So when we talk about, okay, which one do you use?
Dave:I think you should use several of them. Since we last spoke about this, Gemini is basically in the lead now. And if you look at the prediction markets for which one is gonna be in the lead in three months, six months, a year, Gemini's at the top of that list. So if you I I think it's I think it's a mistake to just choose one and go with it. I think you need be keeping your eyes on all of these and especially if you're a developer.
Dave:If you're a developer, man, there's never been a moment in history where you're more powerful as a result of using It's actually incredible.
Michael:What I'd say more powerful and more replaceable if you don't, right? And again, I always bring this example because I think it's perfect because my dad owned an accounting firm his whole life. They were one of the first in our little fishing village to get a computer. And when I showed him AI, he's just completely blown away. And I'm talking about how, you know, existential dread I feel sometimes when I'm seeing like the video generation, how everything and he just, you know, calmly said, he's like, this reminds me of when they launched Microsoft Excel.
Michael:It's like, didn't replace accountants. But as the, you know, the boss who was hiring the accountants very quickly it wasn't, do you know how to put together a ledger? It's how good are you at Excel? And that was the thing that propelled them forward. And I think for the next five, ten, whatever years, it's going to be the same thing.
Michael:How good are you at prompting? I don't care really so much about your programming skills, it's about critical thinking and the ability to prompt. But I like your, you know, talking about your multiple AIs too, because sometimes I pit them against each other. I did that, and it was, I kind of felt mean doing it, even though it was a weird machine. But I went over to Claude one day, and I gave it the chat from chat GPT that I was struggling to have it do, and I'm like, right, chat GPT is sucking here, can you do any better?
Michael:And it was actually kind of mean in its response where it's like, oh yeah, this is why I'm better than this one. You know, so sometimes I guess pitting them against each other can can lead to some funny results as well as, you know, get roadblocks out of their out of your way if you're struggling with them.
Dave:Yeah. Yeah. That's funny. So let let's go into this specific example of what I was doing,
Michael:and Mhmm.
Dave:I'm sure it'll fall into more discussion and give give listeners some some ideas about how to get started. So the product I've released and is working and I've been working on is a product called the Strategy Cruncher. And this is if I have any sort of secret sauce for my trading, this is it. And I've spent a long time deciding whether I was gonna release this into the world because I know how powerful it is. I I know because I've traded with this for twenty years.
Dave:So I don't want to turn this into an infomercial about it, but I think it's good to sort of set the stage about what I was trying to accomplish here and my mindset for doing it. So I knew I knew how powerful this was, this this concept. So it basically allows you to upload a backtest, and it will tell you which rules, which columns are predictive to the strategy. And in a very quick way, it will let you improve your strategy dramatically.
Michael:And I'm just gonna interrupt too because as, right, you built it and you owned it, but you know, I know you don't want to turn into infomercial, but a little bit of a props there. As someone who was able to test it early and has been using it, it has been very, very instrumental in saving a whole lot of time when it comes to some of the strategies that I've been developing, you know, with my service, I recently launched some more strategies that were actually built kind of from the ground up. And with this, and it really helped in just the speed of I'm uploading this thing. This is the base strategy behind it. Here's all the filters and that I think will be associated with it and just having it spitball results.
Michael:Because I think like you mentioned last episode, I'm like the most frequent user because I'm one of those where I'm uploading it and then making a really small tweak and then doing it again in a really small tweak because of the speed of it, I can do that. It's no longer, you know, turning on an optimization on my computer and walking away for two hours while it works and then coming back and getting the data. I'm getting that data really, really quick. So I feel that I'm totally fine doing it over and over and over again, which is very small kind of iterations. And I think that's one of the biggest power behind it is the speed in which it operates and then does iterations.
Dave:Yeah, and it basically allows you to run one backtest and get all your answers for like how to optimize the strategy without curve fitting. So now I showed some early versions of this as I was working on it to some some of the traders that I coach and a couple in particular. One of the guys I showed it to, he's like, oh my gosh. I I can use this. And, like, within a week, he created five strategies that he's trading live.
Dave:He's still trading live, and this was months ago. Another guy so at first so I knew that, okay, I think other traders can use this. And there was a long long while where I thought, man, am I? It's sort of how can I make this simple enough for other traders to really use? The first for the first many years I used this, it was all text based.
Dave:That was just me at the command line writing commands. But I wanted to make it graphical. I wanted to make it show equity curves for the different things that it identifies. And so I went through several iterations and I wasn't I knew that a lot of traders wouldn't be able to use the text interface, but I thought that they would be able to use the charts.
Michael:You were thinking of people like me, right? You're like, Ah, Michael would never understand this.
Dave:Well, likes looking at pictures more than text, even developers like me.
Michael:Well, and we talked about this in a recent episode as well, right? When we were talking about what to optimise for that. At the end of the day, the equity curve is is one of the most important things because it gives you the best feel of how it is that you're going to potentially trade it. So a little bit of, you know, disparaging myself, but I'm glad that it is the way it is because I couldn't couldn't use it if it was super fancy unless we could get AI to tell me how to use it.
Dave:Okay. So I showed it to another one of my traders that I coach, and this guy has been trading for a living for some time now. He's a very good trader. He's got several strategies. I showed it to him in one of our calls, and I get a text from him the next morning.
Dave:He's like, Dave, I couldn't sleep last night thinking about this tool and how I can use it. And that's when I knew, okay, I think I think it's I think I'm onto something here, and I need to release this. Like, the other thing he said was, don't release this. Don't give it to anybody else. So and I know where he's coming from because I thought the same thing.
Dave:Like, why would I give this away? I just wanna use this for myself. But what I realized is I'm I'm kind of on a mission to create the most trading edge I can in the world, and that's through my own trading, but also through the traders I coach and traders I help. And this is totally smack dab in the middle of accomplishing that mission by getting this out to as many traders as I can because I know it's I know it's powerful because it's what I've used for twenty years to trade well.
Michael:Well, yeah, and it's not gonna degrade, I couldn't imagine, any of your strategies because for the people, again, who haven't played with it yet and don't know how it works, you're uploading your own backtest. So it's not like I'm grabbing Dave's strategies and optimizing them and trading them myself. These are these are my ideas, they're my strategies, the bot is simply just telling me which of these filters are going to work better. So for that person who is telling you don't release it, well, don't worry, it's not, you know, we're not looking at what you're doing. We're doing, you know, everyone's doing their completely different thing.
Michael:I'm probably based off what I know about the the people that you deal with and everything, probably one of the only like investors that is using this for really long term things. So yeah, I'm certainly not coming after some of those day trading edges for sure.
Dave:Yeah. So when I set out to do this, I've used several programming languages over the years. I mean, lots of them. The ones I've used most are probably C, PHP, and Perl, and Python now. Done some jobs.
Dave:I've done a lots of lots of different stuff. But what I wanted to do when I created this tool and the web interface for it, I wanted to use this framework for PHP called Laravel. So it's basically a layer on top of PHP that provides a lot of convenient things for PHP developers and for running a PHP based website. Knew this was the I
Michael:understand any of that, but I know a lot of our viewers will, so I'm just gonna nod along like I understand what you're talking about.
Dave:So my only point here is this was a framework that was new to me. I was familiar with it, but I'd never created a project with it. I'd looked at the documentation. It's pretty convoluted, like, just to find a starting point and how to get started with it. A little bit complicated, right?
Dave:Like there's just this mountain of documentation that I would have to read through. But this is like the perfect scenario for ChatGPT. So I just started off saying, Okay, Gatcheepti, I'm creating a project in Laravel. Give me, like, the most basic bare bones demo project for me to create that just says, hello, world. Give me the steps for doing that.
Dave:And, of course, it spit back exactly the steps I needed to do. And very quickly, I had something like a very simple hello, world program that I could use on my machine just very quickly.
Michael:Mhmm.
Dave:So it was ideal for that kind of thing. And for specifically Laravel because there's mountains of documentation that it knows about it and has read through, and it's just way more efficient than doing Google searches because it has the context you're asking for. If you get better at prompting and giving it background about very specific things about what you're doing Think about when you do a Google search, you don't get very specific, right?
Michael:It's Usually, yeah, you're writing like a a sentence if that
Dave:Right. Yeah. And then you're using your logic and stuff to sort of pour through the results, see something that's closed, click on it. Well, with, of course, with The Chechen Boutique, you could write a whole paragraph, I mean multiple paragraphs, and that's what I got started doing. Like I would get I got in the habit of giving it a deeper and deeper prompt, you know, more and more specific.
Dave:And the more specific you get, the better your answer's gonna be. So it was I just got into this workflow for this specific thing where I was new to and it was just it was kind of mind blowing how quickly I was able to get into this routine of very, very quick productivity and accomplishing a lot way more quickly than I could without it. No question.
Michael:It is. Yeah. And then my journey with it's a little bit different because someone who knows nothing about coding. Like, I literally couldn't be doing the things that I'm doing now without it. And it is the first, like, there's a Mac dedicated app, and it is the first app that I open every day.
Michael:And for the same reasons that you're talking about is I, last night I was out burning cardboard, because that's what you do when you live in the middle of the woods, right? Get a lot of cardboard, I get a oil barrel that I burn things in. But I've got in my ears, I had some idea for this this trend following system that I wanna test, and I can just I can hit the go button in in these real test GPT that I've built, and I can just hit the verbal thing. I don't even type to it too much anymore. I go, okay, let's start exploring some things.
Michael:And I'm like, want to build this, and how would I code that into real tests? And it'll give an example. And then I I went back to it with some feedback. And yes, it's that back and forth that I think really changes the game because I'm not just looking for a basic idea. But, you know, that expands out to to everything.
Michael:And the I don't know if you've played with the deep research feature yet. That thing is wild. Like we're looking for a new car. Need a bigger car because we've got a lot of kids. Simply stating, Okay, I'm looking for, you know, a seven seater SUV, and I want you to look at only, you know, rank the ones by how child safe they are.
Michael:And, you know, this is my price range. And this is, you know, reliability, whatever. And then it takes ten, fifteen minutes, and it gives you like an essay of all of these different, cars to look at is is really, really interesting. And the thing that I again, for someone who's not a programmer, I love the democratization of, again, the things that I just could not for me to get where I am systematic trading was I was going to just write off like two years of my life to go in and learning Python and learning all this. But now I don't really need to, right?
Michael:I just need to know enough about backtesting and I need to know enough about system development to be able to smell BS. And I think that's kind of the big skill that when you're using these AIs, you really need to keep in mind is they'll they'll lie to you sometimes. So you need to make sure you're checking you're checking back all the time.
Dave:Yeah, I think that's a there are a couple downsides, but and you see those trotted out and the media stuff and all the time, like hallucinations and such. But there's such a minor annoyance. And it's like, I mean, even the it doesn't take long at all to figure out, okay, maybe it has something wrong here because, like for example, it might suggest a method to use for a response to one of my prompts, And then I go and write in the code and the method's not there. Like, it doesn't work. Was like, well, okay, this is not the end of the world.
Dave:Right. It's you just go back and say, oh, you sure? Like, This is the error message I got there. Why was I getting that? And it'll turn along.
Dave:Yeah, to get upset about hallucinations or write it off because it doesn't code as well as you perhaps. I've seen some developers do that. That is such the wrong attitude and is kind of dangerous. Still hear from it's a guy I play poker with. He works for a company, manages some developers.
Dave:And I was like, yeah. How how much are your developers using ChatGPT? They said, oh, yeah. They're not using it. Like, okay.
Dave:That's a big problem. If you're not if you are not using ChatGPT now and you're a developer Yeah. That's a big problem. It's not gonna take long before I I predict that there's gonna be one person billion dollar companies because of what a very talented developer can do now with ChatGPT. You basically have an army of interns right at your disposal.
Dave:And if you can learn how to use that well, there is a lot of you can accomplish a lot with one person.
Michael:And that's the that's the other thing that I I, you know, we were joking a little bit before we started too, right? There's the whole, you know, almost a pushback of this is going to replace a lot of humans. And this is going to, you know, I like a people that don't want to use it because they're almost afraid of what it is and what the future is. And the way I always look at it is you can't change. It's common, right?
Michael:It's not going to change it. So you might as well use it to be as as superpowered as you are. And I 100% agree with the people building billion dollar companies by themselves, because myself, you know, stats at trading, website and the service that I run, I probably need a couple of assistants to be doing it at the level that I'm doing, and I don't, right? I'm, you know, recording the emails that are getting sent out, they're getting reviewed and proofread and, you know, things like this that, I'm just basic language and grammar, can copy and paste and you say this is what I'm going to write, what do you think of it? And it can give you some ideas.
Michael:All the way up to advanced coding and everything you can do, it's going to be very prevalent, it's going to get much better in the future. So it's like if you're not using it now, I think you're I think you're you're crazy. Have you experimented with any of the custom GPTs yet? Because I know I was going to talk with the real test one. Have you played with those at all?
Dave:A little bit. There's there's actually one we, you know, we've talked about poker out here on this podcast. There's one that there's at least one poker GPT, custom GPT. ChatGPT is very good at doing hand reviews. Yep.
Dave:Sort of like being a poker coach. So it you know, hand reviews with poker players are super fun. There's all this there's a lot of nuances to this and you get a lot of opinions about what you would have done in this situation or not done, or if you made the if you made the right move or how much of how much, how big of a leak do you have because you made that move? JWPG is excellent at doing hand reviews, and the more information you give it, the better it is, and it's it's pretty impressive. So that's another good example of but but their custom GPT is just for that.
Michael:Yeah. I think that I love the coach aspect of it too, right? Because I've seen people who are doing it, discretionary traders who I haven't gotten to yet are still discretionary traders are uploading their journals and everything to it and just asking it questions as opposed to using the trade journals that are out there. I think there's some journals. I think Tradezilla, for example, is using LLM in it now, but there's a couple of these that are rolling out.
Michael:And the idea is that you could just have a conversation and say, as opposed to just looking at the raw data, you could have a conversation. So, okay, well, where am I doing well? Where am I doing poorly? And it will say, Okay, well, you know, at this time of day, your strategies don't do well. Or, you know, I've noticed this particular your symbol and expand that out.
Michael:But the the custom GPTs I really like, because if you're struggling in an area and ChatGPT doesn't know about it, it really helps. So again, you know, you guys can reach out to me because I have that RealTest GPT and we had the interview with the creator of RealTest on it. And they he was nice enough to send here is took all the form data scrubbed any personal information, names, all of that out of it. And I've it has now every question that was asked on the form, and then the answers to it. So not only does it has that have the help guide, it has this as well.
Michael:And I would say half of my GPT messages are just me talking with this very specific one that has gotten very good at learning. I've been uploading also some sample scripts that I have and things like this. So it's really cool. Think when you build a custom one for yourself, that you're kind of tweaking it over time. So it's learning about you and whatever it is that you do and and getting getting better as it goes, as any good assistant should.
Michael:Right? You hire an assistant, maybe they don't know the way that you work, and then over time they get better as time goes on.
Dave:Yeah. So that's really cool. I remember us discussing that in the interview with Marston. Yeah. So that's pretty much a really cool creation that's come directly from a conversation on this podcast.
Dave:That's awesome.
Michael:Yeah. It is it's amazing because it know I don't need to specify that what I'm building, I don't need to specify go to real tests, learn about real tests, blah, blah, because it just the only thing it knows is real tests. So when I talk to and say, I want to build this strategy, it's coding in real tests language and it understands it understands. And I I guess it's it seems to be getting better over time.
Dave:Yeah. Well, it it makes perfect sense. You know, if you, you know, think about doing a Google search and and the type of problem you're trying to solve. If if you're If Google has to search the entire internet, that's one thing, but if you give it any sort of way to focus, obviously it becomes a much easier problem to solve. So yeah, I could totally see in that situation with the RealTest GPT, it should be way faster, it should be way more focused on the problem and this type of solution, so that's great.
Dave:So one thing I wanted to bring up is what it's not good at And you touched on it a little bit there with I'm not optimistic that uploading your journal results to it would get very valuable information. And there's a whole maybe, but I'm a little bit pessimistic on that.
Michael:Why is that? And again, I want to make sure we're specifying. Journal is the way we say journal, which isn't, you know, I was feeling good today, just raw data sheet and say, what do you think of these trades?
Dave:That can work. What I think it does not do well, and what it will not do well for a long time, I think, and that's being creative.
Michael:Mhmm.
Dave:One of the things I put at the bottom of each message I sent out to my newsletter is this is not written with AI. And I don't use AI to create content because it's sort of like it gives this veneer of being creative, but it's not really creative. And with traders, you have to be creative. Trading is a creative endeavor. You have to be thinking of stuff that's unique, and the more unique it is, the better chance it has of being successful.
Dave:So that's for trading ideas, that's for content. So if you ask Chichi Chatuchi Boutique, like, give me a a profitable trading strategy, not gonna do that. Right? It's gonna give you an answer that on the on the surface might seem accurate to a layman, but it's not gonna be unique by definition. Like, think about where it's pulling information.
Dave:It's just basically a glorified Google search. Right? It's that's where it's getting its its information. So it just doesn't make sense to me that it's gonna get to a point where it can be creative. There's probably another word that's better, but that's what that's what I've fallen back on.
Dave:And, you know, so No. And and you can
Michael:see that. Like, when an email comes in and especially the headline is written by AI, that's the one place that it does not do well. If you say, hey, you know, I'm launching a restaurant, give me 10 names for the restaurant, they're all probably going to be awful. Now, they may inspire something in you to say, Oh, know, I kind of like the, know, some of these and I'll combine them or whatever. But yeah, when it comes to pure creativity, but I was just talking more about pure data analytics, right?
Michael:You know, raw, like simple things like, hey, again, this discretionary trader is, you know, maybe you're getting a little tired at two because any trade you take after two kind of sucks. Like, you know, pure data analytics, I do think that there's a bit of power there now, understanding the root cause of what could be causing this. It's not going to do a good job at, but I'm just talking. Yeah, just pure here is here is a bunch of numbers. Tell me which combinations of numbers are good and which combinations of numbers are bad.
Dave:Yeah. I think it's I think it'll be good at that. So it reminds me of my wife, who I've mentioned on this podcast before, she's an English major. She is an artist. She she reads a lot.
Dave:And she hates Chekipy, but she's not a data person. Right? But she you know, if you or an eye can read an email and tell it was written by ChatGPT, she can easily do it and easily identify that, wow, this is terrible writing. It's it's obviously bad. But I think that misses the point a bit because you're like, there's that doesn't mean just because it's bad at this one thing doesn't mean there's this whole other area where you can use it to be a lot better at what you do.
Dave:And that's, I think, the real power. I mean, so when I released this Strategy Cruncher, you know, from the start to the finish of doing that, the coding that I normally would have done without ChatGPT, it saved me 70% of the time, I think. Mean, that's a huge chunk of time that it saved me. Was it blew me away.
Michael:And not just time. Let's equate that to dollars. Let's say you were paying a mid range, you know, junior dev or something. What do you think the price of building that for that amount of hours would would roughly take you?
Dave:Gosh. Yeah. I mean, I don't know. But it's it was so obvious to me that the whatever the number is, it's huge.
Michael:And And then what? The CHI GBT's $20 a month or something like that. That, I think, goes to your point about billion dollar businesses coming out of nowhere is because someone may not have the budget. Like, you could have done it yourself because you understood that, you would have just taken you a long time. There's a lot of people out there that currently don't have ideas that just don't have the ability to do it because, you know, the amount of time to learn the coding and then launch it, you know, you're talking five years before it's ready.
Michael:So that's the biggest thing, I think, where, you know, there's people out there right now who have like coding assistance or, or, you know, something like that, that they don't really need anymore. It's it's more of I can have this build for me if I have the idea, and then from the idea, you can actually turn it into something tangible, and then maybe it goes to a coder to, you know, refine and things like that. But, you know, that that minimum viable product is is way closer than it was just even a little little bit ago.
Dave:Yeah. I think so. And, yeah, it's just I I think people think about using ChatGePutine. I think there's some people probably in our audience that think, okay. Well, I'll just have it create the app.
Dave:Right? Like, it's easy. So I think that it's it's not like it's no work at all. You have to have the smarts and you have to have some vision about what you want. And you have to the more experience you have, the better off it's gonna be because you're gonna you can be prompting it and exactly leading it to exactly what you wanna see because you might know the right way to do it and save some time doing that.
Dave:But, you know, we're a long way from, hey, Chancey Petite, create this awesome app for me and do all the work. Like, there's it's like not like there's no work involved, my point is. And I think that there's a lot of people who have that expectation about it, and then when it doesn't do exactly what they want to see, they're disappointed, but they're really missing the point there.
Michael:Yeah, I think the best use of, if you're going use any AI, is to pick it up and think of an assistant, right? And then, because it's supposed to be the point of it anyway, is, you know, I now have access to somebody who has way more knowledge about different topics than I do, but the where I thought you were going go when you say what ChatGBT is not good at, but I'll go there as well, is you're seeing a lot of these really crappy YouTube videos of ChatGBT built me this trading strategy, and then you'll watch the video, and the guy literally went to ChatGPT and say, Hey, build me a wicked trading strategy. And, know, and then it gave them something, and then that thing back tested well, and they said, Oh, this is amazing. Now I can make like $10 a day doing this. But like what you mentioned, it's a glorified Google search.
Michael:So it went in and found someone probably promoting some crap strategy that doesn't work at all, and then fed that to him. So the the fact that you need knowledge is not going to change the where your knowledge needs to exist needs to change. So for me, ChatGPT just means less time trying to figure out how to code a certain script and more time thinking about things in the market that could generate edge, that I could then back test later. And yes, I can't go to the real test GPT and say, build this and I test it, but it puts together the framework that saves me a ton of time. I throw it in there, I iterate with it a little bit if it's spitting out errors or whatever, and then I can at least go from an idea I had to something that I'm testing to see if it actually has edge in ten minutes as opposed to a couple hours.
Michael:And then doing that over and over again allows me to generate an idea, dispose of it if it's bad, really work on it if it's good. And that initial area just gets way, way faster now.
Dave:Yeah. So let me give two specific kinda hardcore examples to the developers, traders in the crowd that I think would be a useful thing for a ChatGPT project for them that would improve their trading. One is learning and using GitHub. I still wanna do a whole episode on this. GitHub is extremely powerful, and I see a lot of traders using it the wrong way or not using it at all.
Dave:There's a lot of stuff that's a perfect example of what ChatGPT can help can hold your hand and help you get over the hump for using GitHub. It's an incredibly powerful tool for traders managing your code base, using branches to add like a testing code base. And then when you do a pull request to get it into the main branch, have it automatically be deployed to your production environment via GitHub actions. GitHub actions is I've worked with a couple of traders who are doing some amazing things with GitHub actions. That, and it's ChatGPT knows GitHub Actions really well.
Dave:Take a look and like ask ChatGPT, suggest 10 ways that I could use GitHub actions in cool ways that I might not know about. And look at that list that comes back and see how you can use that and see, you know, use your head to figure out, okay, how can I use this in my trading to improve what I'm doing?
Michael:Yeah, we should do that episode soon because I know you've been bugging me about it and it every time I look at it, it scares the daylights out of me as a non So let's just do it as like, I'm going to teach Michael how to use GitHub. I think that'll be a really cool episode. We'll go in, might be one to, to, to really help people that are like me that kind of glossed over there a bit. It's like, I don't I don't know what I'll get have action is, but I had some experience at trade like very rudimentary at trade ideas, but I'll be interested to see that. And then the beauty of of this too, just to go, you know, talking about you mentioned, right, asking ChatGPT about it.
Michael:Things like that don't scare me as much as a non tech person now as they would before. If someone says, Yeah, Hey, I want you to look at that. I'm like, Okay, I'll get a human to show me like just the basics. And I know that if I get stuck somewhere, I don't need to go and bug that human again. I can I can show I have this assistant, right?
Michael:And I think that's another thing that people should really look at is, right, there's a whole world of things that might be off limit before that you can now kind of push into because you have this assistant that can kind of help guide you around and do things. Yeah, I'm way more interested in trying that now that I, know, you gave me that idea of I can just poke chat GBT and say, hey, you know, I'm I'm lost here. What do I do? And hopefully it can it can push me along.
Dave:Yeah. And and get in general, there's like some weird edge cases that you can get yourself into where there's a command you need to be able to run, but you've never run it before or you have run it six months ago and you forgot all about it. That's like the perfect kind of scenario. And it's very helpful for that kind of thing. Here's another very specific example that I used it for that maybe it gives some traders some ideas.
Dave:So I have some, I call them, like, workflow scripts or scripts that I run daily that I that I need to create, and I'm, like, constantly sort of automating certain parts of my trading business. And I I have a certain format from writing, you know, years and years of doing this. I have a certain format that I like to use when I write a script. And the the way I like to do it is I I try to make it as generic as possible. So I've said this many times.
Dave:I'm not sure if I've said it on the podcast or not, but but good developers can predict the future. And what do I mean by that? I mean, you anticipate what users might need and you build that into what you're doing now. Like, build some features into a script so that you don't have to maintain it in the future or you can maintain it a little bit less or you can more easily add a feature that you know is gonna be requested in the future. So one of the things I always do in scripts I write, I write I put a command line option in there, and it's called pretend.
Dave:So sometimes it's a long running script. Sometimes it's a script that interacts with grades or something. I always put a pretend mode. So it runs it actually completes the script, but it doesn't actually do the thing that it was designed to do. It just tells you what it would do if it weren't running in pretend mode.
Michael:Oh, okay. Yep.
Dave:So I can't tell you know, it doesn't come in handy every day, but when you need it, it's really, really handy to have it, and it's good to have that in there right away. So I was writing a Python script, and I couldn't remember exactly how to set up the command line options. Like, what are the what's the right way to do this in Python? Done this in Perl, done this in other languages, but I just didn't know how to do that in Python. So I said, okay, ChatGPT, here's what I here's the command line options I wanna do.
Dave:Here are the defaults that I wanna have. Like, go ahead and give me the right way to do that. And that was a perfect way for it to tell me I was able to communicate because I have this experience in these other languages, and it was able to give me sort of the right way to do it in Python very quickly. So that's another good example of things you could have chatty chatty p two do that can be very helpful to your business. And that's not directly related to trading, but it's it's related to, like, a supporting process that I have that that helps my trading.
Michael:Yeah. And and, again, for me, the the big thing that I'll just poke on is that, you know, don't be afraid. Like, you know, always coming at this as the guy who just doesn't, couldn't get into this world as deep as I am before. The only thing I could play with is graphical interface, back testers and things like that. Is that, you know, now something that you might be looking at and might be trying to avoid because you think it's too daunting.
Michael:It's just way better when you have an assistant that works whenever you want to work and costs $20 a month, even if you end up going the paid route. And, you know, so if you're someone who works full time job and you're listening to this podcast, you're like, Oh, I wish I could get into systematic trading, but I don't know anything about coding. It's like, well, now you can. So when you get home from your job and the kids are asleep or whatever, you can just open up this app and say, Okay, you know, you could start as simply as, you know, give me some of the best back testers to look at. We have an episode on that.
Michael:You should probably listen to that instead, but it will get you started in in a direction. There's no the big, from the non coder side of thing, and Dave saw this a lot when I was working at Trade Ideas, the biggest thing that I would always come into is just being stuck, and just not being able to, you know, there's something that I know that if I can get ahold of a developer who's usually pretty busy, to get me unstuck, I can then work for another four or five hours before I get stuck again. And the problem is everyone's so busy, it's sometimes it's really hard to get unstuck. So you end up not having the tools to figure out a solution to a very basic problem. And that's your real roadblock to do the next thing.
Michael:But now with AI, you could just simply ask it. And that's so the the fear of I'm gonna put all this time and energy into something and not be able to work is is gone for me because I know that whenever I get stuck at a point, I have an assistant that I can just call on and push me through.
Dave:Yeah, totally. So here's one final very quick example unrelated to trading that was super useful. So I was shopping for home insurance. I saw our home insurance was the rates were had increased quite a bit, and I thought I I just I knew I could get a better rate elsewhere. So I took the PDF that described my current insurance rate and what I was getting for that.
Dave:Mhmm. And then I got a quote from another provider. And when you look at those, like, you think, okay, home insurance, what's the big deal? Look at the deductible, the cost. There's like all sorts like there's a whole bunch of stuff that I didn't know.
Dave:So I uploaded it to JackGPT, both of these, and said compare them. It gave me a beautiful comparison in the standardized format and even gave me a suggestion about, Okay, which one is better? It was a very convenient way to do that. And it's not that alone isn't going to change the world, but just to have that very convenient view, looking at these side by side exactly like you'd want to see them, and just the snap of your fingers was just awesome.
Michael:Well, I think that goes to the fear thing before, right? Someone before may might look at that and say, I'm not I can't understand these these crazy contracts, but you don't have to anymore. Right? You can say here are some some back and forth on it. For me, my non trader relating thing, I'll say I already talked about the car thing, that's been great so far, I'm test driving them on Weight loss, right?
Michael:I'm trying to lose a bit of weight, keep my muscle mass. I went into and for like two weeks I just told it in every workout I did, everything that I ate, I was eating way more calories than I thought, and actually way less protein, than I thought, because it it would track these things. And at the end, I just said, give me a spreadsheet. And it gave me a spreadsheet of every single day and and a little table saying this is what you're averaging for protein and fat and carbs and this whole nine yards. And this is one of those things that again was possible before.
Michael:But it's not as easy as just pulling this out of my phone and just saying, hey, I just went to Starbucks and had this. And then it goes to the Starbucks website and it pulls the macros down and puts it. So yeah, trading, amazing. But I would say start playing with these in these outside kind of fringe areas for little little side projects. I guess one one last example before we go, the buddy of mine I was talking about, he had a problem with his car, and they were going to charge him $2 to fix it.
Michael:And he went to chat GBT because it was a computer thing, and it ordered him a little USB stick or something off of Amazon for like $20, and he changed the sensor or whatever it was himself, and the whole project cost him like a hundred bucks. So, yeah, tons and tons of things you can do for sure.
Dave:I love that. That's so great. Yeah. I mean, just as you're talking, I thought of, like, three or four other examples that I'm not gonna go into. We don't want this to go too long.
Dave:But, man, this is great. I'm sure I bet in six months at most, we'll probably be doing another one of these episodes of new stuff. I mean, it's really incredible how quickly things are changing. It's really exciting to me.
Michael:Yeah. So, you know, the main takeaway I'd say for the audience is if you're not, do it right, even if it's random stuff that we just talked about now, it's, it's, it's coming, it's, it's here, and it's going to get more in your face. And what's going to happen is the edge is going to be taken away from you and given to other people. You could, you know, test more strategies or you could be doing things faster, you could be doing things more robustly, you could be there's, you know, and the people who I think just in life in general who don't use it are gonna be kind of left behind by the people who embrace it. So I embrace it.
Michael:And I a % agree. I think in a couple months, there'll probably be something else that's really cool that we'll end up talking about. But I think that was great. And as always, I'm Michael Noss.
Dave:And And I'm Dave May. We'll talk to you next week on Line Your Own Pockets.
Creators and Guests
