Line Your Own Pockets

Michael: 00:00

All right, everyone. Welcome back to another episode of Line Your Own Pockets. Doing a little bit of a follow-up this time as I'm continuing my slog, I guess you could call it my journey into Amibroker to work on more day trading systems and kind of refine those. And periodically we might do little check ins like this, but offline we had some cool conversations about kind of where I was going and how I was feeling. And we thought about just turn on the camera at that point and hit play and and talk about it with you guys.

Dave: 00:32

Yes. I think this would be good because, you know, I worked with a lot of traders to get up to speed with Amibroker. Mhmm. Some of them from scratch. Some of them never done any coding.

Dave: 00:46

And there's definitely a learning curve. There's some right ways and wrong ways to do it. So And so it's been interesting having you start from the ground floor using Amber Broker. I think it was a super interesting episode last time, and I know you've made a lot of good progress in the past couple weeks since we recorded that. So yeah, looking forward to getting into this.

Michael: 01:08

Yeah, and I I'm I'll be peppering you with questions throughout, I'm sure. But first of all, shout out to, you know, your your Ami Broker course. Again, we talked about it that I picked that up. And I think having that base strategy to start with, not only helped me, but actually helped the AI that I'm trying to build at the same time. So that's, I think, a tip right off the bat.

Michael: 01:35

For those who are doing that, I do a lot of custom GBTs where you can, you know, in the back end feed it a bunch of stuff. I do a lot of that, and one of the things that I did is I uploaded that and said, this is, you know, the syntax, this is what we're looking to do, this is how it's written. And I find as soon as you give it a couple scripts like that, it is way better at directing and redirecting questions because it knows the basics anyway, and it's you're like pointing it in the right direction, again, which I think is huge. So yeah, shout out to you for that because that's again, helped me along the way. And then another quick thing just while we're talking about AI is I've specifically programmed this, and I think this is something that people in my area should do where I've told it not just to give me the answer, but to I think the prompt that I use was like a five year old that was dropped on his head.

Michael: 02:28

And I said, I want you to explain everything to me. Like, why are you doing this? What does this mean? What's going on? So when I when I paste the script and I say, okay, let's try, you know, a break of VWAP or something like that.

Michael: 02:42

And it will go through and it will say, this is where I've got the VWAP, and it will like break it out. So I find it really good not only to help me, but also to kind of educate me as I go, because I'm learning from it at the same time, which I think is is great as opposed to just, you know, build something and let it work for you entirely. I think it's great that it makes the notes and it does all that inside the code as well.

Dave: 03:05

Yeah. I mean, I think that's totally great that you're doing it that way. I see a lot of traders try, like, essentially ask too much of AI. And if you don't understand what you're having it do or what you're asking it to do, then you're going to end up with something that's not good. That's really why I created the course and gave the template away because I knew that people were going be using ChatGPT and other LMs to do this, and that is the starting point.

Dave: 03:38

I've heard the same message from other traders that they tried doing AFL before with just LMs. You kind of get yourself in a pretzel dealing with it because it's not very good at just without much context. But as soon as you give it the template, the context, okay, it's here's your starting point. Start from this. It's light years better when you do it that way.

Dave: 04:13

And I love what you were doing, but I love also what you're saying about having it teach you. Like there's one trader I'm coaching now. He's got up to speed on AFL from nothing. He's not a programmer. It's not his background, but he used his template and got up to speed really fast.

Dave: 04:33

He went through, even using the template, when he first started using AI with it, would get because he didn't understand it deeply, he would get in this loop and get sort of spaghetti code out of it. But once you start learning, asking it, okay, what is this line actually doing? And the more you learn about it, the more context you can give it, the better your process and system is going to be for coming up with strategies for an Amber broker, just like you said.

Michael: 05:11

Yeah. And it's really important that you know enough to know when it's lying. And that's that's basically where I want to get to. I never want to learn necessarily all of the ins and outs. I think a lot of that is kind of unnecessary as these AIs get better and better.

Michael: 05:31

But you need to know enough to do a bit of like bullshit detection where you have to go in and say, okay, that doesn't make sense. And sometimes that is just done with the results. You get a line that's shooting straight up and you go, okay, something got broke here. But it's important to be able to go through and say, no, you know, that's not what I wanted or something like that. And it was funny, it was the same thing when I built my RealTest GPT, where it kept trying to give me Amibroker type code.

Michael: 05:57

And I could tell because the semicolon at the end of every line, it's something you do in Amibroker that you don't do in RealTest. And then as soon as I fed it some of the things that I had already built in real tests, it worked really, really well. So I think that's like a huge number one tip. If you're struggling, just get any script, even if you grab it online, if you, however you grab it, that is a working Amy broker script and feed that to it and say, no, this is right. If it doesn't look like this, it's not right.

Michael: 06:24

And it does a really good job. Because I think it's like tries to default to Python or whatever it's doing it from. It doesn't, sometimes I find this thing gets lazy. It doesn't wanna go and do the work. It wants to just feed you something that I found online or something like that.

Dave: 06:41

Well, as part of our interactions with you learning this, I actually created something that I've released with Makekit that I'll talk about in a minute. I want Let to me give you an example to illustrate how you should be using an LLM and how you shouldn't. So imagine you had access to the smartest person in the world, and they could do your bidding and would 100 percent follow your instructions. Well, you would still have to give them tons of context with what you were asking them to do. Mhmm.

Dave: 07:25

Without that, it's pretty much worthless. Like they're not gonna be able to help you that much. And it's the same thing with LLibs. If you don't give them a lot of context about exactly what you want, you're going to end up in this loop of just back and forth and you're going to end up saying, Well, L and M suck, or maybe I'm just missing something. So the more context you give it, the better it's going to be.

Dave: 07:54

And like I said, that's why I specifically designed this template and it's way better. It's by far the best one you're ever going to find because unlike every other AFL strategy that you might find on the internet, this is one that supports the process for adding columns to the backtest, which is the only way you're ever going to end up with a strategy that's actually profitable. So it shows you a good example of that. And yeah, it's a great starting point.

Michael: 08:32

Well, and that was another big thing, right? Is that I have in RealTest, this library of columns that I keep playing with and adding to and all of that, and that was big. But yeah, it is not nearly as simple to do it in Amity Broker. So it, you know, it certainly took a little bit more work to get all that together. But again, it's like anything I've really noticed with this is that the first step is the hardest, right?

Michael: 09:02

Just getting through the door. And because like a lot of things I would imagine, I'm not a programmer with code is that once you have something that's a baseline of it, you never start from scratch again, right? You always can grab little bits and areas that you need and copy that file in and get rid of all the stuff you don't care about and start again from there. And doing that has been huge because again, as soon as I got a very simple strategy working, working in, you know, I can run a back test and it comes out and it seems to make sense what it's doing. Then it's just modifying that.

Michael: 09:38

And from the, you know, copying and pasting that into the LLM and say, we're making these changes, go do that. To just being able to always go back to something that works if I break something, those two things are great. Because there's just sometimes where I'll just put, you know, the code, your example code and what I'm working on side by side, if something is breaking and I can compare to and do things like that. And I think I made the example last time was if you're gonna write something, the best thing you can do is open up a Google Doc and just throw a bunch of stuff in there to start, you know, stream of consciousness typing. Because for me, that was one of the biggest things is just that the initial, right, get the data in, run a simple test, and then, from there, that feels to me anyway, like it's like 80% of the work done.

Michael: 10:23

And now it's just iteration and, you know, changing things and updating things from there. But for anyone who's kind of nervous to get going, I would say just get going and get going with like the simplest kind of basic strategy that you could possibly think of, even if you have no intention of ever trading it, because it will it will get that initial kind of fear out of the way, I think.

Dave: 10:45

Yeah. Well, the the the strategy template that comes with the course, it is a working strategy. And in fact, it's a great starting point because I've got a couple traders that are going live with a version of that strategy, like uses literally the exact one as a starting point. So if you have ideas and you need a strategy, that's a great place to start. So there's I spaghetti code earlier, and that concept comes from software development.

Dave: 11:28

If you work too quickly, you try to get something working in your code, you've got to accomplish something, you try to do it the quickest way possible. A lot of times you take shortcuts. Do the quickest thing just so it works even though it might not be maintainable. If you do that a lot, you end up with what's called spaghetti code. Technically, works, but it's not very extensible.

Dave: 11:54

You go back and read it and it's hard to understand. That's what you want to try to avoid. You end up doing that with your use of LLMs if you don't understand what you're doing well enough. You'll end up with And think the modern We used to call it spaghetti cutter. I think what you call it now is AI slop is, I think, the modern term for it.

Dave: 12:17

But it's very easy to get into that loop. Mean, imagine if you had it's sort of like an overeager intern that's completely overconfident and willing to do whatever you want. Imagine if you're the CEO and you have an intern and it's overeager and it's overconfident and you rely too much on that, things are going to fall apart really fast. So you really need to be quite discerning as you use these. And the more discerning you are, the better you're going to be.

Dave: 12:57

That and it involves, you know, really understanding what you're doing.

Michael: 13:01

Yeah. And that's where I'm still not, but that's where that's where I'm working to get at. And I do think the instruction of, you know, teach me kind of as you go, I think is is huge and understanding that I'm coming from a different backtesting software. So at least there's some underpin of that as well. And that's still the kind of parallel I'm always baking in my brain is, oh, that's how Amity Broker does this that RealTest does this way.

Michael: 13:32

And trying to kind of line those up, I think really helps me. And for people outside of that, you know, even if you know Excel and things like that, I always think Excel is a good kind of starting point for, you know, basics when it comes to coding and the way Amity Broker seems to handle data and things like that are very, very kind of Excel like. And then also starting with something simple, right? And this is where, you know, we're gonna talk about the, what I'm trying to do with Amnibroker, right? Where I have these swing trading strategies and they're working great.

Michael: 14:07

And, you know, last year was a great year and all that. And so I understand a lot about it's generally simple strategies that work the best. Like this is why we're starting there. And I think a lot of people when they get into this, they go right into the deep end. They're looking for kind of the craziest setup or something they saw online or something like that, where probably just for the better of their trading, but certainly better for learning a new system or learning to code even from scratch if you're not there, starting with just something that's super basic allows for a couple things.

Michael: 14:45

One is the code shouldn't be that complicated if the strategy is very basic. And then also the ability to check on the chart for what's happening is just vital because you want to run the back tests and see the results and grab the trades and you want to go check those and make sure they're doing what they're doing. And if your strategy right off the bat has a thousand indicators and all this kind of complexity, it gets very hard to do that spot check. And spot check for me has been huge because, you know, the first couple of times I did it, what it outputted, regardless of what the equity curve looked like, wasn't what I was going after. It wasn't the core of what I wanted to see.

Dave: 15:25

Yeah. You're totally right. And, you know, when I created the template, I thought I was trying to think, okay, what sort of strategy should I have here as a starting point? Should it be something that works well? I was trying to figure out, okay, what makes the most sense here?

Dave: 15:43

So I chose basically an opening range breakout strategy. And I like that one because it's super simple. The other reason I like it is because most people think it doesn't work or it stopped working a long time ago. But as a couple of traders that I'm working with are finding edge with it and going live with it and making money. So think it's a really good example of a starting point that is super simple, but the simple ideas are the good ones.

Dave: 16:16

That's what you can do.

Michael: 16:17

And it's funny that you say that you that you think people think it doesn't work where I kind of look at that and I say, and maybe this is, you know, my technical analyst side or my swing trader side of things that it that it always will have to work in in in some respects, right? It's the idea of the reason the the turtle trading system or trend following works in markets is because for a stock to go from $5 to $10, it's it's gotta break $6 at some point. So the question is when you're buying it at $6, what are all the other rules associated with it? So that's why, and it was funny because I had I had mentioned that I want that's where I'm going to start is an opening range break strategy, because of its simplicity, but also because I believe that this is something that will in some respects for some number of stocks always work until the end of time. And then you just kind of laughed because that was the I didn't know this at the time, but that was the strategy that you include.

Michael: 17:12

So it's kind of funny that, you know, it's both of us from different areas of the world or different areas of trading just come to this simple fact that, right, you're looking for a breakout in something. The question is what? And then, you know, all the accompanying rules after that thing.

Dave: 17:30

Yeah. Yeah. I mean, if a stock is gonna go up and trend during the day, it's gonna have an opening range breakout just by definition. Right? So Right.

Dave: 17:38

They have to exist. But, yeah, it's there's I've heard so many traders over the years say, you know, such and such idea stopped working. I love it because I know that that tells me that they don't have a good process for improving what they're doing. It's usually the kind of trader that I love working with because they are smarter than they think they are. They're closer than they think they are to real success.

Dave: 18:13

And also they've had some periods of profitability before, so they kind of know what they're getting into. That kind of person I I hit home runs hit home runs with, and I I I love working with that kind of person. Well, because by definition identify them really well.

Michael: 18:32

Yeah. By definition for something to stop working, right, it had to have worked at some at some period of time. So yeah, and that's going to be and just like we talked about, and I know we had a disagreement with it, but I'm all about strategy diversification, right? So that was another kind of aspect to it is that I have some intraday strategies right now that I like and I've built with RealTest, but because of the nature of the way RealTest works using daily candles only, they are all by definition kind of mean reverting, right? You're looking for, you know, an extension above and beyond because the way RealTest has to work in order to to work on daily charts, you're either kind of simulating setting out a limit order or stop limit order before the market opens, right?

Michael: 19:20

Because it can know that it went through that price and there's your entry, but it didn't quite go to that price, so it didn't get stopped out, and then here's your close, right? So that's by definition what it wants to use. And I'm actually, you know, really happy with those because they're consistent and they're they're fairly slow and, you know, it's just you're you're looking for a stock to really tank and you buy it and then you sell at the end of the day. So the simple area of saying, okay, I don't want to do that. So I didn't want to start the this day trading journey using intraday bars, getting anywhere close to what I'm doing currently.

Michael: 19:57

And this is, you know, we had the whole, if you missed, we had all a video on portfolio level back testing. This was my simple way to avoid needing to do that right away. Because I just know that this is going to be completely uncorrelated to that. I don't really need I to can after the strategy is done, but it's not a concern for me. And that's my plan, think for the first three to four strategies for intraday trading that I have lined up, is all of them are going to be separated enough that this, which I think is kind of a limitation to Amibroker, I just won't get to because I just know intuitively that they're going to be completely different and it's just not going to be a problem.

Dave: 20:36

I think there's I think I felt more about our disagreement on that episode. I think that's true. Like, the intraday strategies are going to be more uncorrelated almost by default. So you don't really have to worry about it as much. So I think that's a huge bonus.

Dave: 20:53

I think that's what you're going to find.

Michael: 20:55

But right, if I was running, so again, my real test strategies are mean reversion nature. So if I jumped and also done a mean reverting strategy in nature as well, I could see more correlation there. And in in maybe not in most markets, but like say the market's really dumping, know, it's another twenty twenty scenario or something like that. I don't want just more strategies looking to buy the dip, you know, but I know so you can kind of bucket strategies completely differently. Like an opening range breakout is kind of by definition some type of momentum strategy.

Michael: 21:29

So that is just gonna be so vastly different in a completely different universe using one minute candles as opposed to daily candles. I do think that, you know, I can get fairly far in this process before I'll need to worry about the the correlation nature of it. And then by then, hopefully, I'll be enough of a pro and amni broker that I'll be able to whip up something that that solves that problem. It's kind of kicking the can down the road, I guess, is the best way to put it.

Dave: 21:56

Yeah. So So you're implying there is that you're creating this first strategy in but really what you want to do and what you're on the path to doing is creating multiple strategies with Amity And that's specifically why I designed this template to do It's for exactly that purpose. And so here's how I think about it. There's a really good analogy with software development where if you've got code, a lot of it's going to be shared among projects. If you're creating a new software project, a lot of the code that you did in software project A, you're going to be able to use in software project B.

Dave: 22:50

So way you do it as a novice is, well, you just copy and paste.

Michael: 22:58

That was my point.

Dave: 22:59

But that's not the right way to do it. The better way to do this, and you really leverage your work by doing this, is creating what's called a library. I call the column library a library specifically for this analogy. In software, you've got this whole bunch of common functions that you wanna do. When you create a library for it and then reference the library with each new project, you're not copying and pasting code.

Dave: 23:27

So why is that a big deal? Well, let's say you find a bug in the common code that has the common functions. Well, if you weren't using a library, you'd literally have to go make the same change in the code in all your projects. And that's error prone, it's a big pain in the butt. If you've got a library, you make the one change in the library and all your other projects are updated automatically.

Dave: 24:01

That's a huge way to leverage what you're doing. And that's the same way I created this template and the column library, because what you're doing with your column library, it's the building blocks for creating a strategy that is actually profitable. And you want to be able to reuse that, not just copy and paste. I did that for years with the column library until I figured out this process of a way to independently add columns to your library and then go and automatically add them to all the strategies that you're back testing. That is such a huge breakthrough and it's very similar and software developers will understand this to using a library versus just copy and pasting code.

Michael: 24:50

Yeah. And this is, again, I have something experienced with RealTest where, right, you can do, you can have that, that and push it over. And that's, I think one of the things that made it more, I guess, open to me to be able to go do is that I understand that once I get The hurdle for me isn't the process of building strategies, because I have that, right? The hurdle for me is the actual coding side of things. And this is where I think it's really important to use LLMs and to use, you know, mentorships and courses and things like that, is because everyone's going to come at this and they're going to have a completely different sticking point, right?

Michael: 25:31

There's someone out there listening who is a coder, and he's a great coder, and he'll just beat the pants off me building these things. But he just doesn't have the market experience in order to generate the strategy. So for me being able to walk in and I know the first three strategies I want to build, we can talk about them here, don't mind sharing them at all, but I know that going in, so my initial framework is set. So I it's just a pure, how do I brute force that way from from the ideas that I have into strategies that I'm then deploying, and you know, knowing that in some in some orientation, everything I'm thinking about will work, right? I just, I know that the question is, it's a process thing of getting them to work, not whether or not they will work.

Michael: 26:23

And then that is going to be way different from someone else, right? Who's, you know, really experienced at coding and wants to get into this, but might not have that kind of market mechanic side of things. So that, and then also the, I guess, fear I had of day trading as a whole, because I've just been bad at it for so long. And that was alleviated just me doing my systematic swing trading for enough, right? I was like, Oh, well, I am now comfortable letting robots kind of handle it and do all that stuff for it.

Michael: 26:55

So it should be no different from day trading, right? It shouldn't have anything to do with my skill, it should have to do with just hit, you know, program it in and away you go. So yeah, look at, just spend a lot of time looking at where your sticking point is, and kind of ignore everything else for now, and then just kind of brute force your way through that sticking point. So that's what I'm doing with Amibrokers. I am just brute forcing my way into, and I know I'm probably doing everything wrong and I'm breaking everything all along the way.

Michael: 27:21

But for me, that's a lot of the fun, right? Is that I know eventually I will get to this point where I will be deploying one strategy. The goal is within this month, it's the January 2 for those who are wondering when we're recording this. So by the January, I want to have one strategy in development, trading real money, very, very small. And I'm just going to brute force my way through it.

Michael: 27:47

So it's kind of like, yeah, just get to that point where you know what's what you're lacking and just push through that.

Dave: 27:54

Well, it's been interesting seeing you, you know, pepper me with questions as you're going through this

Michael: 27:59

Mhmm.

Dave: 27:59

And showing me say a couple of light bulb moments for me were were the things that you told me chat GPT said that when you asked it, that I knew was completely wrong. So actually, as part of this process, I've gone and added something new to MADE Kit, which is an AFL code generator. And this took everything from the template plus a lot of the context that I realized that LLMs would make mistakes at and basically send you down the wrong path. All that is built into this generator. You can use this form to create a very tight starting point using the template just from typing in common language that just describing the move you want to capture, and it'll come up with a very good starting point.

Dave: 28:58

I'm really excited about this because people have already told me it's really freaking awesome and I know it can get significantly better. I'm getting feedback from them about, okay, here's the input I used for it. Here's the strategy I got back and I can see some little tweaks to make. I think this is going to be a really excellent process and way to create that first starting point that does basically all the work for you.

Michael: 29:29

Well, I think this is why just side note, but relevant where, you know, had a call from the CMT. So I do a lot of work with the CMT, for those who don't know, it's the chartered market technicians, I'm a charter holder for years now. And they're just way more they're like this year, we want to get you on all these podcasts. And they've just been peppering me and I so I called this the president, I guess it is at the time, and was like, so what's what's going on? I hadn't really talked to you guys in a while, and now all of a sudden.

Michael: 29:58

And he said that the amount of interest that they're getting in systematic trading and programmatic trading and all this stuff has just absolutely exploded with the growth of LMs, and he's kind of contributing it entirely to that where a lot of the people who had done, we'll call it traditional technical analysis before, are looking to either automate what they're doing, or, you know, get away from what they're doing and automate things completely differently, and just haven't had the ability because getting, especially in the finance space, getting a quant and hiring them and getting them to work is hundreds of thousands of dollars a year, which is unattainable for even most, know, IRAs and small firms. And now they can do it with $20 a month, right? If they have the right, you know, process of building a strategy or doing some sort of analysis, they can start to build more and kind of go from there. So it's interesting, think on a broader note, I'm kind of a canary in the coal mine for the trading industry as a whole changing as someone who is has done this process, maybe just a little bit before everyone else.

Michael: 31:07

But I think there's a whole wave of people who are just, why am I sitting here staring at the screen watching these waiting for these two lines to cross, when I could test build it, test it, automate it and go from And I also think it's going to make the trading game way more competitive, You're going to have way less, I would hope way less, I'm trying to say this nicely, but losers out there who are just, you know, I saw this TikTok and now I'm going to trade like this and away I go. I think you're going to get way, way more smart people out there who are able to put their ideas into practice with this kind of LLM as a bridge between the two.

Dave: 31:43

Yeah. I think think that I'm not worried about our edges going away anytime soon. Let me just say that. Because I think that the way most people are using LLMs is very naive and they're way overconfident. So I think there's probably going to be more edge than ever up for grabs over the next few years, just because the feeling of overconfidence that people have it and the way that these things are marketed and the way people think about them is, okay, you can chat GBT, give me the most profitable strategy and code it up.

Dave: 32:28

And that's just never going to work, but you ask a lot of traders and that's that's what they're trying to do. So I think there's gonna be a lot of I think there's gonna be a lot of overconfident naive traders coming into the game.

Michael: 32:42

So just I'm gonna hit you with a couple of Amity broker questions while I got you before before we end it. How do I make it not take forever to run a backtest? And this is the this is the biggest gripe that I have currently is that I know the amount of data is different, but I can run, you know, twenty five years in a backtest, in real tests it'll take thirty seconds. Regardless of whether or not, in Amibroker I select I wanted to backtest a month, or I wanted to backtest five years, it takes exactly one hour to run the to run the backtest. And so I feel like there's something that I'm doing that's, again, probably someone else in the audience is doing as well, but it's like regardless, the only way I can shrink it down is if I pick a single symbol to run, which is not ideal, but I can't vary the amount of time because my idea was, okay, well I'll start running back tests on just the current year and that will take hopefully one fifth of an hour.

Michael: 33:40

So it's twenty minutes, it's still long time for me, but better, and then go from there. But it's always one hour regardless of what I put in.

Dave: 33:50

Yeah, it needs to go through all the bars to do the backtest. So here's what I would suggest. If you're running into that a lot, there's a couple different ways to do it. So if you use an IQ feed, it's very simple to just create a new database with however many bars you want to. So you could create a secondary backtesting database with just as much smaller set of bars in it and use that.

Dave: 34:17

The other way is you can filter your backtest to just run on a certain set of symbols. And in fact, you could even do this, run your entire backtest the version, export all those symbols from that backtest to a watch list. And then if you run it again and only focus on that watch list, that's gonna be just way faster. Because it's only gonna be looking at a certain subset of the symbols. So that's a couple different ways to do it.

Dave: 34:50

But you're right. I mean, it's fundamentally different just because you're working with so much more data. And I would argue that it's a good thing that it takes longer because that fact alone is weeding out your trading competition because there's not gonna be it's gonna take a more technically savvy kind of person to even get to the point where you are now. So that's, in my mind, that's a good thing.

Michael: 35:24

Yeah, well, and it's also, it makes you, I think way more purposeful about Yeah. Running each individual because they get in real tests, I can run a back test and I know if it's if I've done something wrong, it will just be immediately evident and I can go from there. So just, probably take as much time probably on on each Yeah. Just because I'm I'm just hitting I'm hitting the back test button just all the time because there's no downside to doing it. Right.

Michael: 35:53

Yeah, what I I like the watch list idea, and again, I'm gonna get IQ feed here soon. I think that idea is good as well. And funny enough, that's the same answer ChatGPT gave me. So I wanted to see it saying, it looked at the code, it's saying it's going through every bar, even though you're selecting. And it gave me all kinds of things to change within the code prevent that.

Michael: 36:13

And I'm like, I don't, I don't want to do that. Right? I don't want to go in there and at this stage and do it, but that's a good idea because yeah, that's, know, we did this whole back testing video and I think it was a great one talking about the difference, but we know one of the things I loved about trade ideas, even though I don't think that back tester is good enough to take a strategy necessarily super live with the limited window, but the speed of it, right? So you could get a very good idea of whether or not your, because for me, first part of the back test is just I want to go through the charts and see if it's doing what it's doing, which is another thing that it could just, I get all this probably user error, but I find the ability to find the arrows for the buy and the sell on the Amibroker charts just way harder than real tests. I don't know if there's something custom that can be done there, but when I right click and it says show actual trade area, it's supposed to zoom to that area of the chart, and it's just not for me.

Michael: 37:12

I don't know if there's a setting or something I have wrong. Yeah.

Dave: 37:15

Don't know. There is a setting where, you know, you can do the right click and say show trade arrows. There's a newer setting in recent versions where it'll When you click on the trade itself, you don't even have to right click and say show, it's just gonna automatically shift the chart and show exactly where it is. And you can change the scaling. Like a lot of times you'll have a whole bunch of bars on the chart and it is kind of hard to see.

Dave: 37:43

So maybe there's a huge range you're showing. But if you if you zoom in a little bit, it should show you exactly where the the entry for the the signal is.

Michael: 37:55

Yeah. Because that's just been the the biggest barrier for me is just that getting getting started. And then because I am on Amity Broker six, and I know there's seven out there, but I'm just it's I'm not cheap. I'm just terrified to change anything at this point while

Dave: 38:11

I'm Well, so so here's one thing about I I have such respect for Tomas and what he's done with Amber Broker because I've seen the changes he's made over the years. I think I started using, like, version three something or four something. Mhmm. Maybe even earlier than that. I've seen the updates he makes over time.

Dave: 38:30

The betas he puts out are usually rock solid. I don't think I've ever had a problem. Maybe one time have I ever had a problem upgrading to the latest even beta release just because I know he takes this stuff very seriously and he does a really good job of handling updates. Is very exciting to see an update because Alf, he has some really good release notes and I'll go see exactly what he's done. And it's very interesting to follow along about, okay, what features did he add?

Dave: 39:10

Ah, that makes sense. So it's very good reading to go through and look at those release notes and see what's actually been done.

Michael: 39:17

Yeah. And that's so that's good because, yeah, I will update at some point. I've just been terrified to, you know, change the playground. I'm just learning to play too much and have the buttons go differently. But yeah, other than that, I feel like I'm already to the point and what's probably what been a couple weeks, I'd say two, three weeks that I'm I'm to the point where I'm starting to really get to the point where I'm going to start using the cruncher and start automate or optimizing things and adding those optimizations.

Michael: 39:47

And then just getting to the point where I'm going to, I'm going to start trading them live. So I'm, again, the goal is by the end of the month, I think I'm just gonna pick a prop account for this to get started with. So I can spend a relatively low amount of money and just, if something explodes, it's not my money to do it in, but I think I'm really close to that point. So it's great. It's and just make sure that, you know, if you're listening to me that you understand that I can't code anything ever.

Michael: 40:15

So, right, if I can do it, you guys can do it too. And I think you guys should be encouraged to do it. I'm sure periodically we'll kind of catch up with my progress here to see if I'm now a pro day trader. I'm certainly not gonna move away from the swing trading Dave, but I'm excited to kinda add a skill or add more strategies to my belt here with this.

Dave: 40:40

Well, since we've been doing this podcast, you've shared some of your strategies and your results and such with me. Yeah, I mean, you've come a long way since we've been doing this. And I can tell that the strategies you're coming up with now, I can tell you got more ideas, you got better ideas, you're trying different things, you've really started using the Cruncher in a way that makes a lot of sense and you're getting great results from it. So yeah, kudos the work you're putting in. It shows.

Michael: 41:15

I appreciate it for now. Like it's one of the things I like about kind of this podcast, this dynamic is that we're both coming from this kind of completely differently. Like I spent a lot more time on the discretionary side than you did, and probably will look back to regret doing it sooner. But I think it's good too for the audience to see that kind of transition live as somebody who is not is not lacking from the analysis side of things or the market knowledge side of things, but is lacking from the other side of things, which I think there's a lot of people out there, know, we talk a lot about short term trading, but even long term investors and things like this could probably find ways to systematize and automate their process. You know, I could imagine someone now who used to read a lot of income statements and balance sheets and is looking for weird ratios and stuff to make decisions off of.

Michael: 42:08

They should even be looking to do a lot of the stuff that we're doing where you're systematizing that, you know, the data gets exported and LMs read it and that kind of stuff. Because I guarantee you that's happening in the hedge fund and the the larger fund world where they're not traders, they're investing, but I bet you they're using a lot of these processes now. Now that doing it has become way easier again with LMs and all of that. So, you know, if you're someone who's stumbled upon this and you're not even trading what you're investing, I think you take all the stuff we're talking about and you just expand that timeframe just a little bit further.

Dave: 42:44

Yeah. So any other Amber broker questions you have while

Michael: 42:48

we're on? Not now, but I have a few. Wait, actually, I had a note. We'll just look at this live. So I had every back test takes an hour, so I already went for those as well.

Michael: 42:58

Uh-huh. Now, the rest I have for yeah. I have for your kind of side project, but we'll do that offline. But we got to everything. I again, I'm just excited to keep pushing forward and I got some good ideas.

Michael: 43:11

I really liked that idea of exporting that Symbolist and going from there to do backtests because if I can shrink the backtest from taking an hour to taking like ten minutes or even a couple minutes, that's going to make my life, especially in this time where I don't really know what I'm doing. So I have to go and break something before it tells me what to fix. I think that'll make that huge. So I think there's a couple of great takeaways for people

Dave: 43:36

who So are following up as one other thing you could do is, maybe your list of symbols from the backtest is still pretty large. What you could do is take a subset of that. You could even do the symbols that have shown up the most in the back in that back test, take like 10 of them, put those in a watch list, that's gonna be super fast to run a back test on. And you have a lot of really good examples, trades to look at to make sure it's it's working well. So yeah, that's another way to do it too.

Michael: 44:09

Yeah. And you just to reiterate to people, we're not doing this for the strategy. It's not like I wanna just trade those symbols, but what we're talking about is that the more you limit the symbols, like if you go and you just pick one symbol, you can do a backtest for five years on one minute data and it'll take you a minute. So it's trying to get to that point where you're getting enough data that you're understanding that what it is you're trying to do is getting expressed well by the backtester, And then you'll just open it up to everything and hit a button and go to the gym or something like that and come back and it'll be done. It'll be done after Yeah,

Dave: 44:42

love what you said about having to be more intentional when a back test takes longer. I think that's totally right. I mean, the backtests that take just an instant amount of time, like trade ideas and real tests, it kind of promotes bad habits because you have that ability, right? So when it takes longer, you have to take a step back and really think about, okay, do I have my back test set up with everything I need to make sure I can get the right answers I need once it completes? Yeah.

Dave: 45:15

And when you think about that in that way, it sort of forces you to think about your process and improve it where you might, you sort of gloss over it if the back test is fast. Ironically, it's promoting bad habits essentially.

Michael: 45:32

Yeah. Because right again, I have my my strategies that I wanna run and I'm working on the first one. And I know every time I hit the button to see whether or not the strategy does well, it's gonna take an hour. So I might as well, you know, take an hour beforehand and just really make sure that I've included all the changes I want to make and do all that. And then, know, it's not bad I guess too to hit the button and say, well, I have to not focus on on this for a little bit of time.

Michael: 45:59

Me take a or dog for a do something like that as well. But yeah, very intentional, but this was great. And again, if you guys are interested in more of this journey, let us know. And what we'll do is probably periodically check up on, you know, how I'm doing and what roadblocks and changes I made. Because right off the bat, I've got a list here from this one where I'm gonna change the IQ feed and the symbol lists and all that kind of stuff we talked about, but it was great and I'm glad that I can grab Dave here live and bug him for all these questions.

Michael: 46:32

But as always, I'm Michael Noss.

Dave: 46:34

And I'm Dave May. Talk to you next week on Line Your Own Pockets.