You're Probably Backtesting Wrong
All right, everyone, welcome back to another episode of Line Your Own Pockets. In this one, we are actually going start a little bit of a mini back testing series, but more about workflows than anything else. Think about what, know, proper ways to actually to go through and to hammer out a process, because just like in trading, I think your testing process is very important so that you can just replicate it over time. And this ties into a couple of our most recent episodes where we're talking about a new product that Dave's working on and is launching because now he has a time loop by the time these episodes come out to help us with this. So we're going to start a little bit, I think, with how most people try to figure out what filters or indicators or aspects actually affect their backtests and some kind of incorrect way people do it, and then some ways that we think make a little bit more sense.
Michael:So first of all, how do you see everyone do everything wrong, Dave? Because we talked with this a little off stage, and that's why I make the joke, because I've seen everyone do as well.
Dave:It's you know what? Backtesting is so powerful. And when you first start doing it, you're like, Man, I can do anything. Like, I could test any idea. Like, I've got the whole trading world at my fingertips here.
Dave:But what most people do almost all traders, when they start backtesting, do it wrong. And so what do I mean by that? So here's what they do. And this is probably what a lot of listeners do too. So I'm really excited about this because it's so eye opening for people when they do it the right way.
Dave:So here, let's just describe the wrong way. You code up a backtest, you got an idea, you code it up, you run the backtest and you know, you've got all your code in there for all your thresholds that you think are going to work. Like, let's say you're trading, you've got a trading signal and you're looking at relative volume, and you think if it's above a certain level that those trades are going to be good. So you put that in your code, relative volume above two, say. Then you run your backtest, you look at the results, and you're like, Man, that didn't turn out exactly like I hoped.
Dave:Let me try, maybe it's relative volume three, maybe it needs to be higher, right? There's too many trades in that first run. So you go back and you change your code. You run another backtest, you get a completely different set of trades, and you compare to the original. That's not exactly right.
Dave:So you go and do another adjustment, like maybe it's like distance from fifty day moving average, or whether it's above or below. So you add that. Yeah. And you run another back test. So you've already just with you've already run three back tests.
Dave:And that's what it doesn't sound like. It's, it sounds, powerful. It sounds like they, this is what you're supposed to be doing, but that is so wasteful and inefficient that process. There's a completely different and better way to do it. And once you see it, it's you kind of can't unsee it and it saves you a bunch of time.
Michael:Well, and what I think gets people, and this is absolute personal experience, is because you've already shrunken from discretionary to run doing it this way. You've already shrunken the amount of time it takes to come up with these ideas by like one one millionth, right? From going through a bunch of charts and trying to say, Okay, I would have taken a trade here. I would have done like the complete old school way that takes hours and hours and days and days. And then what you described takes ten minutes, maybe a half hour or so if it, you know, the back test is really long and takes a lot of computation.
Michael:So I think the reason that people kind of stop there is because they're like, holy smokes, I just saved the infinite amount of time, can now do so much more. And then they don't realize is that if they just took one step further, they save even more. Like there's the whole I'm thinking of, you know, those parabola type graphs that even off at a certain area. It's the same thing where you've only taken one leap, you can take another one. There's eventually a point where going further doesn't really help, but a lot of people think they've reached it as soon as they start backtesting and you there's two or three other other jumps to make in order to make that that more efficient.
Michael:I think that and I think one thing that is, I don't know if this is just like our monkey brain, but when you hit the back test button and the chart actually starts to fill out in front of you, I think there's something weirdly satisfying about that that people like to see. But the main thing I think is that, people are like, Wow, I've just saved so much time, things can't get any better than this. But, you know, Dave actually was the one that turned me on to the kind of this new way of thinking about things, And it does very much take that next leap. So why don't you describe that a little more?
Dave:Yeah. So the way I the best analogy I can come up with for this is, let's say you're trying to shoot a bull's eye at like 400 meters away. So you've got a target and you've got to like, what sort of tool would you use to hit that target 400 meters away? Right? You might think, okay, I can use a slingshot.
Dave:Well, that's not going to get anywhere close to the target, right? So most people, when they back test, it's like they're using a pistol to shoot at a target 400 meters away. And you think you're getting close because the bullet's heading there, it's heading that distance, but it's not going to be at all accurate. If you hit the target, it's going to be purely by accident. And so what you really need is a sniper rifle, a well calibrated sniper rifle.
Dave:Take some time to set up a shot. You know, put the target in the sights, maybe take one or two stabs, and you'll be able to hit the target every time. But most people, when they're backtesting, they're using a pistol and they're wondering why they can't find a strategy. So this workflow that we're getting ready to describe allows you to use a sniper rifle to hit that target rather than a handgun. So what do you think about that analogy?
Dave:I think that's it's the best one I can come up with. Maybe you can come up with something better.
Michael:No, I think that one works well because it just shows the leap in, if you didn't have a slingshot and someone handed it to you, you're so impressed. And that was the same thing is that I didn't have the ability to back test. When I got my hand my first back test, I was so impressed. You don't keep going and it almost it's kind of human nature. Someone gives you something that's way cooler than you had, you don't instantly grab a hold of it and go, Yeah, okay, what else do you got?
Michael:You go, This is so cool. I want to play with this tool for a while. So there's not that instinct to go, and what's next? Right? And what's next?
Michael:And what's next? Because all like I mentioned, you've gone, you now have something that may have taken you two days before that you're doing in in a half hour max. And it's so much better. But it's the whole kind of the saying that I like to think of is right is all you have is a hammer. Everything looks like a nail.
Michael:So like, Okay, I have this tool. And I think part of it, the only thing that would kind of shake people out of that is, A, if someone comes along like us and hopefully tells them there's a different way to do it, or B, you've just been doing it for years and you get bored of you get acclimated to the amount of time you're saving, and then you start to think what's next. But it shows that there's a bunch of different tools. And I know it's one of those that if we record this episode in a couple of years from now, maybe we even think of thought of slightly more improvements from here. Right.
Michael:So part of it is you're always trying to improve your trading setups, but also the other part is you're trying to improve your process to discover new trading setups. So I think that's where we're where we're going with this.
Dave:Yeah. So the right way to do it is think about your backtest a little bit differently. So your very first backtest is not going to be the one you're not going to take all those trades. You're not going to be trading everything that comes up in that first backtest. Your first backtest is collecting a dataset and you want it to be comprehensive.
Dave:Yep. Knowing that you're probably only going to end up with the final version of your strategy taking about, I don't know, 10%, some small percentage of the trades in that initial backtest. But it's important to have that first backtest have a comprehensive list of all the instances of your trade signal. And that's the key starting point and the key part of the process that most people miss because they're trying to create a backtest that goes a lot of steps ahead. Okay, I'm going come up with something right now and that's going be the final version and I'm going trade everything that's in it.
Dave:And that just, that's, that's the, this guess and check approach, which you get into, like we just said, like three back tests right away.
Michael:Yeah.
Dave:You get in this mode where you're in this rabbit hole. You always feel like you're real close and you're just one step away, but you're really a long way away from having something that actually works and has trading edge.
Michael:Unless you're real lucky, I think would be the caveat to it. Because you could come up with a off the cup kind of combination of things that that look great. It's just going to be like a once in a lifetime thing, right? So that's why we're talking about processes. You have to have a process from that base all the way in.
Michael:And the reason I really love this process as well is because I find not only the refining process, I think really speeds things up, but the fact that you're thinking about this is kind of a broad test gets you to the point where you decide whether your basic idea has any merit at all. Right, so if I go and I back test something simple as, hey, what happens if I buy stocks that are making new all time highs? And then, you know, hold them for a month and sell. Just something basic like that, that's my basic idea. A lot of people will also add in a whole bunch of indicators like you're talking about with this kind of guess and check method, but the way you look at it now before you go into the refinements that we'll talk about, is I'm just testing that concept as a very broad ideal.
Michael:And I can do that, and if the equity curve is flat and it doesn't do anything, go, that just doesn't have edge. There's just no reason for me to waste my time to to go and explore it any further. But you've you can do that testing very broadly. And then just you got to think of your basic statistics. If your first backtest only has a small amount of trades in it, it is less just mathematically robust than if you start with something that has thousands of trades or or tens of thousands of trades or hundreds of thousands of trades.
Michael:You start with that, then you have a way better understanding of, okay, what I'm doing is profitable over the long run. And now we can start to kind of carve it out and get a little bit a little bit deeper into the nitty gritty from there, as opposed to just wasting your time. Because I've seen that a lot, too, where people haven't done that broad testing. So they're trying to kind of put lipstick on a pig, right? They're trying to refine an idea that in its broadest sense just doesn't work.
Dave:Yeah. I like that you mentioned that you have a large number of trades to start with. I mean, that is so important because let's say you do get lucky with your guess and check approach, which sometimes happens, right?
Michael:Yep. Just random luck.
Dave:Yep. Now, even when that happens and you get lucky, you're still going to be trading a strategy that could have more trades in it. If you start with a broad base, you're giving yourself the best chance to come up with something that has a large number of profitable trades in it. When you take that guess and check approach, just by definition, you're going to find some subset that is going to be a smaller number of trades, less profitable than if you start from this broader base and work your way down, eliminating the worst trades and basically maximizing total profit of the strategy you end up with. It's a very powerful way, and it's a way to know that you're going to end up with something that's optimal when you use this approach.
Michael:Well, it's kind of the only way to limit risk while also maximizing gain is to have more trades because you're, you know, it's a basic kind of form of diversification. If there's, you know, if during a day I could take one trade or I could take Trent ten, I'm with one tenth the size across those tens, I'm going be better off taking 10 because net net, I'm probably A) going to make more money because I'm exploiting an edge more often than compounding or whatever you're looking at. But at the very least, I'm going to take less risk and have a smoother return. Because as opposed to right day one comes, you take one trade, you either make or lose money and then day two comes, it turns a binary event into right much more finite binary events. So instead of you make $100 or, you know, you lose $50 that day, well, any combination of those things between that zone with 10 smaller trades, you can you can end up doing.
Michael:So, yeah, I find just more trades is going to lead to a smoother equity curve. It's going to make it easier to trade that strategy. You know, when you have a losing streak, it's going to be probably a little bit less. It's it's diversification, but not the old boomer way where you're buying, you know, sixtyforty stocks and bonds, but more trades smaller is always going to be favorable, especially in a world that we live in today where commissions, at least in The US, aren't $10 a trade. There's still some still some banks up here trying to charge us that.
Michael:But yeah, so more trades basically just trying to explain that more trades are always better.
Dave:Yeah. Mean, you know, we've talked a lot about a path to confidence. What's your path to confidence to trading a strategy with bigger size and meaningful size, that's the right way to do it. I mean, let's take two strategies, one that you said one trade a day or even less. Let's stick to an extreme.
Dave:Let's say there's one trade per month in this strategy and you have a lot of, let's say you have a lot of confidence in the one trade. Well, it only comes by every month. You better be ready for it. And compare that to something that trades 10 times a day.
Michael:Are you
Dave:going to ultimately have more confidence in?
Michael:Well, and I didn't even think about the missing it, right side of things. I just thought, you know, I was just assuming fills, but you're right. That's another aspect as well. If there's 10 trades that day, you're going to be way more okay missing one than if there's one trade that day and you miss it. Yeah, that's something I didn't even factor.
Dave:Yeah, basically it basically reduces the importance of any one trade. The more you take. Right. And
Michael:there's big psychological benefits there, too, right? You know, if you're if you're someone who's especially if you're watching it in real time and you have kind of impulse control problems, if you miss that one trade and that one trade, especially if it was a banger, like you would have made like $23 with that trade, You're going to be way more likely to do something dumb, like take a random trade or try to chase that or do something like that. Then again, if you're like, okay, well, I've got 10 trades here. I missed that one. That one would have won, but it sucks.
Michael:But because it was a smaller win and you've got all these other trades to make up for it, you're probably going to be less on tilt if nothing else.
Dave:Yeah. So let's go back and so I think we've established here how most traders backtest wrong. Let's talk about the right way to do it. So you have an idea, and let's say it's trading gapping stocks, let's say, in some way, and you're going to go long gapping stocks in a certain situation. Well, the first thing, again, your first backtest should be collecting a large trade set, not the ones you're going to ultimately you're not going to trade this entire set, but you want this to be comprehensive.
Dave:So what does that mean? Well, you want to basically do a backtest that captures the signal with very little filtering at all. So you want to capture every instance of the signal that you can and use that as your starting point. Like I said, you're not going to trade everything that comes through that first backtest, but that's going to be your starting point from which you can use a process to whittle down and come up with the optimal trade set for you to trade for this strategy.
Michael:Yeah. And I always the example that I gave when I was teaching people about it is kind of, you know, you've got an idea in your head, say you're a sculptor, it's like clay, right? You're not going to go and start like carving out the fine details right away. You put the clay down, you get a general kind of outline. And I'd say that's like step one.
Michael:And then you start getting you get into it after that. Yeah.
Dave:So back to our example, let's say you have a theory about how you can improve the system or a critical filter in the system that's going to make it more profitable. We talked about relative volume. So let's say you have a theory, okay, the higher the relative volume, the more likely these trades are going to be profitable. So instead of going and putting that as a rule in your strategy where you say, okay, only take the trade if relative volume is above, say, two, the critical step that you're going to take here is don't modify the rule set for your strategy, but only what you want to do is add a column to your trade set that has the value of relative volume that you can make, that is the value that you can know prior to the trade and add that to your Trade Set. So now you've got, you haven't modified your Trade Set at all.
Dave:You've got the same trades in your Backtest. The only thing you've done is added a column for Relative Volume. Now you run that Backtest, you export it to Excel, and now you have a column for relative volume, and you can bring that up in Excel, you can filter, you can figure out the exact point you should be using to filter out the bad trades and leave in the good trades. So
Michael:Well, and I just hit on that a little bit. It sounds simple, but it's a kind of direct 180 in thinking, right? So one is just a binary thought, right? Above this good, below this bad. Then what you're thinking is more of a continuous idea of, you know, you could chart it, right?
Michael:That's one thing, you could chart out like a bell curve of distribution of returns, things like that. So although it seems like a very minor difference, it was a big moment for I'm sure me and other traders as well, where you're now thinking, Okay, I need to completely flip on the head from this binary, good, bad, on, off, whatever, and just think more continuously in nature. And then that flows, I think, into a lot of the other stuff that you're going to chat about, too.
Dave:Yeah. So now you've got the one back test with an additional column. Now let's say you come up with another theory, like above or below the fifty day moving average, the common one that traders think about. So the next step would be add another column that describes that situation. Now, as you alluded to, some people will just say, Okay, is it above or below, either one or zero?
Michael:And
Dave:as you say, reduces the problem to a classification problem. That gets rid of a lot of details about that situation that you're going to want to know about. I mean, you know, if you're like way below the fifty day or just a little bit below the fifty day. There's a big difference there. You don't want to reduce that to a zero when some of them are real close, but just under some of them are long way away.
Dave:And Capturing the distance from the fifty day is a way better, a far better way to describe that situation in a way that makes a lot of sense.
Michael:Yeah, and you're being one of my favorites. I'm a chartist by trade, right, a CMT, and one of my favorite sayings for technical analysts is basically you draw your areas with crayon, not a pencil. And it's just basically saying that, you know, the classical support resistance and chart patterns, these are all kind of fuzzy, And most people are totally fine with that, but then when it comes to numeric values of is it above or below this? All of a sudden they get really staunch on, right, number above good, number below bad. But basically what you're saying is, say your hypothesis is correct, and it being above the fifty day moving average is good.
Michael:Does that mean if it's 1p below the fifty day moving average, it's it's now shit? What if it's 10ยข below? What if it's right? So just by doing this, you're at the very least saying, Okay, let me give a range. Then it also answers a lot of questions when people are testing things about, well, what about the 51 period moving average or the 49 period moving average?
Michael:And the fact that you're now you're no longer thinking binary, but you're thinking it tenuously, it opens up bands to everything. Now you're just basically saying, instead of this this kind of mythical moving average that you're throwing on there, you're just saying this is a proxy for some sort of trend or something like that. So now let me look at that as my proxy for trend and say, you know, hey, if it's too far below it, it's bad. If it's too far above it, maybe it's also bad. But within a certain zone, it makes way more sense.
Dave:Yeah. So I remember the reason I started doing it this way. So way back in probably 02/2008, '2 thousand and '5, somewhere there, when I first started day trading, I was following a system that I found on a blog and that I started communicating with the trader that was doing the blog and basically following the system. And luckily it worked well right away. I was making money right away from it.
Dave:And I remember asking him, like, I think it was about relative volume, like, I saw this trade come up today. Did you take it? And he was like, No, I didn't take that trade. I said, Oh, well, why didn't you take it? And he said, Well, I don't take trades under relative volume two or whatever it was, two or three, like it was some number.
Dave:And I was like, Oh, really? Well, how'd you come up with that threshold? And he's like, Ah, you know, it just felt right. You know, it was sort of the, like the fuzziest answer you could imagine to like somebody, a data nerd like us. You're like, well, the first thing I think is how do you know that's the right threshold?
Dave:There's probably more trades you can add to your system if you chose an optimal value for that and had a good way to do it. That was just the ultimate guess and check, where I needed a process for myself to say, Okay, what's the right value to use there? And in a way that I can have some confidence in it. If you come up with something that's just completely arbitrary, even if it works, you still don't have a path to confidence unless you know what the thought process was for coming up with it. And then you can have a process for believing in that.
Dave:It's just a better process to fully understand your strategy and have a path to confidence so you can trade it with size and you can trade through a drawdown when it comes because you know the strategy well, and this is a way to do that.
Michael:Well, I think it also allows for more nuanced answers to nuanced questions, as opposed to so if you go back to the old way, the simple thinking is, right, above two of all good, below two hour of all bad, right? It's a very simple question with a very simple answer. But if you are doing it in kind of this new way, the question is more, how does ARVOL affect my trading strategy? And you can get a lot more nuanced answers with that. So, for example, what you would like to see to make sure that that is even a robust filter at all is that around a certain number, it does okay.
Michael:What? Because if, you know, if it's you have these bursts of high performance randomly throughout that curve, it might not even be predictive at all. And that's just an assumption that you've had. And maybe just by cutting off a certain amount of trades that have some other factor affecting them, the equity curve looks better. But you're you kind of come across the right answer by asking the wrong question, if that makes sense, where if you're asking in this continuous way, it's very robust where, again, in theory, what you would hope is that as the ARVOL increases, at least to some metric, your performance increases in some way.
Michael:By thinking continuously like that, can say, Okay, it makes sense. And then where you decide to cut that off could, you know, is going to vary the equity curve some, but you're just not picking just a hard number out of the sand. Again, for us stats nerd, the fact that he said two ARVOL and 1.9, not 1.97 or 1.94 or whatever, just shows that there's a lot of kind of human bias in there. Cause you pick two because it's round number, right? So you pick two because it's it's double normal volume.
Michael:Yeah. And that was basically the probably the reason he chose it. If it was a more nuanced number, it would probably not be that simple. It's like what they did an exercise I saw once just in universities where they got people, they just put down a piece of paper and they gave them a sharpie. They said, Okay, I want you to put random dots on the page.
Michael:And every single one of them had a pretty even distribution of dots on the page, which is not random at all, right? It's you're specifically there's an empty space there, let me put in a Whereas if it was random, there'd be clusters that, you know, didn't make sense, and large blank areas. So yeah, the fact that it was two showed that it was very was a very human systematically picked number.
Dave:Yeah. Yeah. Yeah, I like the way you were putting that with I really like the way you put that. And the thing it reminds me of is once you start adding more columns to your backtest, then it's almost like you're setting up a competition between the filters. Which ones like all these aren't going to have the same effect on the profitability of your strategy, right?
Dave:Each of them are going to have a different effect on the strategy, and some of them might not have any effect at all. And what you're setting up here in your backtest is a way to answer all those questions. Like how does this column affect profitability? How does this other column affect profitability? How does this indicator affect the profitability of the strategy?
Dave:You're setting up a way to answer all those questions with one backtest. Just a single backtest, and you can answer all questions that you might have simply by adding a column. And as you get into this process, a couple things happen. Might not sound the way we've described it. It might not sound that much easier because you're having to add a column every time, but what you can do, you end up having a standard library of columns for yourself that you can add to every backtest, and you can do that easily.
Dave:So every column you add not only helps your backtest that you're running right now, but literally every backtest you will ever run into the future too. So what you're doing is you're creating a process where the original idea, the main signal, the main impetus for your strategy doesn't have to be incredibly unique and powerful. Like all of a sudden you've got a path to and a process to basically improve any trading idea, any trading signal, and see and basically come up with a profitable strategy from it, which is incredibly powerful. You cannot do that with the guess and check approach that most people take.
Michael:And it's funny because you end up finding a lot of commonalities, I find too, when you do this. If bring the same filter set to every backtest, eventually you start to learn things from there as well, right? You ran 10 backtests and you find out that the same filter set seems to be having a good effect on 10 backtests on 10 different systems, well then that opens up a whole another door about why, and then is there a system to be built just based off of this thing? And I laughed while you're talking there because I feel like quite often people, again, come to systematic trading because they say, okay. If I have a robot do the trade for me, I have way less work to do.
Michael:And they listen to this podcast, and then we end up giving them, like, a couple hours at least or days more work to do. But this one is, you know, is very super worthwhile because it it just shows you that, you know, there is an infinite amount of things that you can do to make your trading better that is not trading, which is why you should have, you know, systems do your trading for you. But I just love that having that universe of filters really does help because you'll have your go to. And then just like you may find ones that are new and interesting, you'll also find ones that are a complete waste of time. Just the inverse.
Michael:If you test 10 different strategies and something that you thought was a really good predictor never shows up substantive or interesting, then it's probably something you can just take out and say, well, no, this isn't something that's good for at least the kind of trading that I do.
Dave:Yeah. And it's, you know, this is for me, it was not just a difference in, saving a couple hours. This is the difference between the ability to create a profitable trading strategy and not. Like, you can't do it. Like, you literally can't do it with the guess and check approach.
Dave:It's not possible. But with this approach, there's always a path to creating a new strategy. There's always a path to improving a strategy you already trade. This workflow is just light years ahead of what I was doing before, and like this is there's just no way that you should be doing anything different. Backtesting, you should be doing it this way.
Dave:There's no question.
Michael:Well, and I think people naturally, and again, when I say people, I usually mean myself too, naturally gravitate towards this, even if they don't know it, by refining their guess and check approach. So, you know, for example, you use a fifty day moving average, right? So what they may do is they say, Okay, I added the fifty day moving average, the back test looks a little bit better. So now let me try, you know, a 51 period or 55. So they end up doing the same.
Michael:They guess until they hit something, then they try iterations and nuances of that. So they're kind of doing the same thing. They're just doing it in a slower way. Right? They they they're doing the the idea of like a homing missile.
Michael:Right? It makes micro adjustments as it gets closer and closer and closer. So they're doing the same thing, but they could have just started off by throwing this column in and doing a test themselves in some sort of way so that they could have known that outcome to begin with as opposed to, you know, a little bit, okay, I put a moving average that's too big and that's throwing off, so let's do one that's too small and vice versa. And yeah, it makes a lot of sense. And it's one of those that I think people eventually will get to the right answer.
Michael:But just being explained to it could save months or years to eventually pull them into that answer.
Dave:Yeah, I think that there's you're not going to come to the right answer no matter how many times you guess and check for a lot of situations.
Michael:And no, mean, will come to the right filter answer eventually, but it will just take it. Will just take a thousand, right? Your battleship might be a good example, right? You're randomly the old school board game battleship, where you're throwing out darts and then one hits, and then you're playing around that general area until you hit the target. Whereas if you could just see the other side of the board, you just saved yourself a countless amount of attempts.
Michael:Right?
Dave:I love that example. It's been a long time since I thought about that game, but you're told that's exactly right. That's a great analogy. Like playing battleship the way you're supposed to play it versus yours. It's like essentially cheating, like just looking at the board and and going right on the right.
Dave:I
Michael:know we're all right. I ran one test to tell it told me where all the ships are and now I just yeah, put in the right numbers. So that's the way I looked at it, because I always found that, you know, what we're going talk about is the system that you're building now, the amount of times it lined up with what I was thinking, but then took the amount of guessing and moving back and forth or running iterations of or optimizations against a certain strategy away because it just yeah, it just told me. So I thought it was great.
Dave:Yeah. Yeah. So I love that analogy. I'm going to have to use that one versus the bull's eye from 400 meters analogy.
Michael:That's a great one. Just a pretty beard. Make sure you write that.
Dave:Alright. Well, I think that might be a good place to stop the
Michael:Well, we can't we can't leave you hanging. Right? Or are we gonna we do another little teaser for what you're building at the very least.
Dave:So, yeah, that's
Michael:kind of the just so people know, right? And I'm forcing Dave to do this. It's just the the iteration of of what we're talking about is what we talked about in the last episodes where we're basically the Dave has been building a system to really help this process along. But it was important that I think that a lot of people listening to this understand why it's so important to think this way as opposed to the old way so that they can understand why it's so much better to do this filter check as opposed to not.
Dave:Yeah. So the product I've released is the Strategy Cruncher, and it basically allows you to upload a backtest and basically tell you where all the battleships are. I love that analogy. So I'm going use it. That's perfect.
Dave:Yeah. So if you're interested, you can go to davemabe.com. There's a Strategy Cruncher link at the top. And this is, like I said, this is a thing that I've used for twenty years to create strategies for myself. I know it works because I spent years saying, I'm never going to give this away to anybody.
Dave:It's too valuable. And I finally figured out that that's not the right approach. And like I said last time, I'm on a mission to create the most trading edge I can in the world for myself and for other traders. And I see so many traders using the guess and check approach in a way that they're never going to get the right answer. So that's why I released the Strategy Cruncher.
Dave:That's why I'm offering it. Everybody that's used it so far has been blown away. Mean, your experience has been I think you've still used it the most. And yeah, I can tell that your I can tell that your eyes have been opened for a lot of the strategies that you trade and you're coming up with new strategies and ideas as a result of using it.
Michael:Yeah, I'm looking at it, you know, and, you know, it's it's kind of like the AI, where I'm looking at it as like an assistant, where the way I'll do things is I have my own ideas of potentially what could affect the base strategy that I come up with. So the same thing. I come up with a basic trading strategy, you know, it's this old, slow swing trading stuff that you hate so much, but I'll come up with an idea for it, and then I'll start to write down and have some ideas of what I think should affect it one way or another. And then I put it in the system, and then when it agrees, that's great because, you know, I thought the same thing as that. So, okay, let's put that in and go from there.
Michael:When it doesn't agree, that again just leads a separate decision tree of thinking about why didn't it agree, why didn't it look at my filters? Sometimes I try to like poke it and prod it to look at the filters and sometimes it just never shows up. Then it's like, okay, well, maybe my thinking behind this is wrong. And then going and doing another another check. And just like anything, it's good because it's completely outside.
Michael:It doesn't know what the trading strategy is. It doesn't care about any of these things. So, if I think, for example, just use the fifty day moving average that we've used the whole time, if I think that the fifty day moving average should be up, sloping up, and it should be above it, and it's telling me, no, all your your crappy trades happen when that's happening. Well, that's like a bit existential, right? Because now I need to go and think, okay, this whole concept I thought behind the trading strategy may be completely wrong.
Michael:So then I take the dogs for a walk in the woods and I try to think about it and then come back and and do it again. So I like it. I like anything that is will unbiasedly tell me sometimes that I'm wrong. So then that really makes you think about, okay, why and potentially what can I do to change it? Or is it telling me something that I can then go explore with another trading strategy altogether?
Dave:Yeah. So that phenomenon where, you know, you're sure that something you add to a backtest is going to be the most important filter and then it doesn't show up on the list or it shows up in the report and it's telling you the opposite of what you thought, those are the big moments that really open your eyes to how to improve your existing strategy. But that's not really the most valuable part of it. The most valuable part is taking what you just learned and figuring out how to create a new strategy based on what it's just told you. And it's those things that this tool will tell you that you would never find out about otherwise.
Dave:It might take you years to discover it and it'll tell you really quickly and point you in the right direction really quickly. Again, I love the battleship. It's going to tell you where the battleships are. That's that's I love that.
Michael:It's perfect. And, you know, I know we glossed over the whole building your own filter set idea behind that. But don't worry, we're going to circle back to that because I think that that is not only something that if you agree with everything we said in this podcast, that will help you out dramatically, but it will be one of those very front loaded investments of time and energy in which once you once you do and have good, then if for every backtest you do forever, when you do backtests, you'll have these to kind of carry with you and you'll be kind of ever growing with them, which again, just to give everyone a whole bunch more work is like one thing that I'm looking to do now is I've gotten a lot of back tests the way I want them. And I feel like there's a natural kind of break in my workflow. So I'm going to go back and refine my filter set and add a few to it.
Michael:But then that opens up going back to all of the strategies and then going through them again just to see if these other filter sets rise to the top of of importance. So, you know, we're going to go over that. I think that will be part of this kind of mini course we're doing as well very soon because I think that will be very, very important to give people a place to get started to build this kind of base of ideas. So, yeah, anything to to add before we close it down?
Dave:I think that's a good place to stop. I like like, we'll we'll come back to there's so much we could talk about specific columns
Michael:and how to
Dave:your library. That's let's let's do it next time because we could talk forever about that.
Michael:That's really where we could, I think for this topic, probably just turn on the microphone and go for like five hours. So we're very intentionally trying to, you know, break these up into little bits. They're they're very consumable for you guys. And as always, right, to make sure you're not missing any of these, make sure wherever you're seeing this or listening to it, you follow, subscribe, all that kind of stuff. Then hit us up wherever you can find if you like this content as well.
Michael:And as always, I'm Michael Noss.
Dave:And I'm Dave May. Join us next week on Line Your Own Pockets.
Creators and Guests
