Line Your Own Pockets

Michael: 00:00

Hello, everyone. Welcome to another episode of Line Your Own Pockets. Got another good one from a user question about signals when it comes to building systems. But I think it's better probably just to hear it from the listener and or themselves. So we'll just get Dave to to read the email.

Dave: 00:17

Yeah. This is a good question from Matt. He's a long time subscriber to the newsletter, long time listener. So sent me this really I'm just gonna read the whole thing here. So I saw your video about signal strength.

Dave: 00:30

I think he's referred to my YouTube channel where I put some videos. And how good how a good signal should have positive expectancy before any filtering or optimizing. I thought it'd be good for a great podcast. How to tell if they're if it's a good signal or not? If possible, at least a guess before running a long back test, and he's got five suggestions here.

Dave: 00:55

Chart pattern, volume condition, time of day, general pre filters to exclude stocks you would never take a trade on, event triggers like earnings or new all time high. I know you won't give me specifics, but just a process or way of thinking about signals would be helpful. But then information information that a hobbyist could start to recognize them and gain confidence in whether they or not they are recognizing them. So I think it's a good I think it's a really good question, and it's a it's a good topic that I've been talking about with a whole variety of different traders, traders that are just getting started, traders that have been trading for a long time, like an SMB. So it's a it's a it's a it's a topic that a lot of people think about, and, yeah, I think it's good good to discuss.

Dave: 01:46

So my first question for you, Michael, is, like, what do you think about the question? But also, how do you think about this in general? Like, the signal, the strength for the signal.

Michael: 01:56

Well, and I think it's it's it's a great way to start thinking about it, because you're asking the right question when it comes to wasting your time. You know, it'd be wildly unfortunate to, you know, spend a whole lot of time thinking that whatever ever signal you're using for your setup is actually predictive, and you've just found, you know, the whole correlation does not equal causation thing. Right? I I don't know how many times I've seen people say, oh, know, I I figured out this new trading signal or or whatever. And then you look at their back test and they basically just recreated any trend following model.

Michael: 02:35

Right? So you know, it's like, oh, I found this cool pattern or something for when stocks break out the highs or or whatever ends up being. And you look at their back test and it looks very similar to any old school, you know, buy a new all time high, buy a new breakout of a trend, buy new something. So that's the first of all, I think the question is great because that's the thing that you're trying to figure out is whether or not the signal itself is predictive or whether or not it's just you're getting lumped in with not random, but just things that exist over time. Like, you know, do gap ups fade on average 50% of the time and the signal you found also fades the gap 50% of the time, and and you really haven't found anything there.

Michael: 03:21

So that's it's a good thing to look for because, like we always mention, is that the amount of ideas that you're gonna come across in your career are unlimited, eventually, once you get into it. Finding making sure you're focusing your time in the right area, I think, is is the question, is the place to take a look at. Mhmm.

Dave: 03:41

So how do you think about the the base strategy before you run it through the Strategy Cruncher. Do you look at that and, like, does the equity curve at that point matter any to you? Like, how much do you

Michael: 03:57

how much weight

Dave: 03:58

do put at that?

Michael: 03:59

Yeah. I'd I I don't and I I could be convinced that this is wrong, but I think it'd be hard. I want to make sure that there is some predictive value in just the base thing, you could call it that. So you take the strategy and you make it as simple as you possibly can and see if there's any sort of predictive value. Right?

Michael: 04:23

So does you know, let's say you're kind of fading gaps, just for a random example. If you're trying to detect whether or not the size of the gap matters, you should be able to very kind of quickly pull that out just naked, kind of on its own. Right? And and ask yourself a basic question of, you know, do a bar chart of the probability that a gap's gonna fill on one axis and and the the, you know, the the magnitude of the gap on the other axis. And if it's random data, then I probably just kinda walk away and don't bother looking at it.

Michael: 05:00

It's like, first, I wanted to have that signal prove itself completely on its own. No stop losses or profit targets or, you know, end of day exits, or if it's a a swing trade, you know, you'll pick a a maybe three days out, five days out, a month out, you know, a couple random type intervals and then test that as well. And only if it survives then, that's when it's it's kind of warranted my time. If not, I put it back on the shelf. I don't necessarily discard it entirely, but I put it back on the shelf and say, well, unless I can think of something that will prove to me that the signal is predictive and it's not all the stuff I've done around it, then it's just not it's not worth my time.

Dave: 05:40

So what about I mean, why don't you just run that through the cruncher and like have it tell you which things are important and make the curve look good?

Michael: 05:52

Well, it's it's I think you'll you'll like the quote on why, but it's the whole confidence thing. Right? You need to you need to have confidence that the core idea is one that makes sense, because your idea should always come from some, I don't call it behavioral finance or whatever you want to call it, some some core idea of, you know, I'm not just buying something because this line crosses this line. I'm buying it because I think there's something that happens to humans or, you know, our stop loss is getting hit or people, you know, chasing news or or what is the thing that I think is occurring in the real world that creates some amount of edge. Then from there, I find the job of of the cruncher or or people who are doing optimization themselves.

Michael: 06:36

The job of optimization is to try to find out under conditions it works best as opposed to trying to find under what conditions it it works kind of period. So it doesn't have to be a great edge. It's not gonna be an amazing edge when you're testing it raw, but just any amount over over some sort of baseline. And I think that baseline is important. We can talk about that later.

Dave: 06:59

Yeah. I think I think we mostly agree there. I think that here's the way I think about it. If your signal is not super strong, then you're gonna have to like, the filters you apply to it are gonna have to be they're gonna be doing a lot of the work. So there's like, the the the part that's gonna be creating your trading edge is the filters you apply to it.

Dave: 07:33

So you can that you you can do that. Like, you can like, I call it sometimes, you know, putting lipstick on a pig. When you start start out with an equity curve that looks bad and or not great, and then you run it through the crusher, and it's gonna tell you things you can do to improve it. Right? But the when you when you start with something that goes down or is not that great, your job is a little bit more difficult when you're trying to decide what rules to apply, and making sure that you're not curve fitting, making sure that the ones you select are actually predictive.

Dave: 08:11

It it just makes that part a little bit harder. So Well, go ahead.

Michael: 08:15

That point so so my kind of rebuttal to that point is that point, if you're fine like, so you have a back test. Let's just say it's not very positive and it's not very negative. Because if it's very negative, that opens up a whole other conversation of maybe the inverse is true or something like that. But it's it's random nonsense kind of noise. Right?

Michael: 08:35

The equity curve just muddles around zero for a long period of time. And then you put it in the cruncher, and the cruncher finds all these filters that look great. At what point is the filter now the signal? I don't know if this is like getting too semantic, but Yeah. So for example, I I was using gap ups as and and the size of the gap up, do I think that's important?

Michael: 08:59

Well, if the cruncher comes to me and says, well, here's a bunch of filters that are are indicative of relative volume. Right? Well, is now relative volume of the filter. So it's not like it's a useless exercise. It could open up a very interesting question.

Michael: 09:13

But if there was random noise before the filter and there's a good trend after the filter, it's like, did the thing that I put in it make any difference? Right? Should I now go back and maybe my base strategy should be relative volume based? And if that ends up being interesting, then I crunch that data as opposed to the the data I was initially looking at.

Dave: 09:36

Yeah. No. I I think that's a great point, and it's it's a bit of a philosophical question. Right? And the way I think about these is when you apply a filter like that, and then all of sudden it kinda feels like a different starting point, I I think of those as universe filters.

Dave: 09:59

So when I come up with a starting point, it's all it's it's really like a certain universe with an entry tactic bolted onto it. So the entry tactic might be fairly generic, but then, you know, one broad based predictive filter applied to it can be a different starting point. So you're right. It's it's a bit of a esoteric philosophical discussion, like, what is the starting point and what is, you know, a filter for optimizing. But, yeah, I think that I think you're spot on that when you notice something like that, it almost it it makes you have a different starting point, and maybe that points to that starting point, but also an opposite starting point.

Dave: 10:56

Like, the ones you removed, maybe you take those as a mirror opposite of what you did when you applied that filter. Right? Like, so say, in your example, fading a gap, doesn't look great, but you apply a filter and you're like, okay, Relative volume looks a certain way for those. Getting rid of a bunch of trades, and you have a different starting point using relative volume. Maybe you have the ones you got rid of, and you trade those the opposite way.

Dave: 11:29

Now you have two strategies, two different starting points. So it's so I think the it is a bit it is a bit esoteric about, okay, what is the signal? And when you combine that with a universe filter, it it becomes a different thing and a different starting point and a different strategy in itself.

Michael: 11:54

Well, and also, so what what might be interesting here is is a talk about robustness testing. Right? Where a simple way to do like, say you're using a moving average for your strategy, right, and the two hundred day moving average or whatever it is. A simple way to do kind of robustness testing in back testing is, if it works for the two hundred day moving average and completely falls apart for the hundred and ninety nine day moving average, then it's it's probably a useless kind of filter. So what I would do is look at it, again, just using the gap fade example.

Michael: 12:27

Let's say you were doing, I don't know, an opening range break or something like that for your your entry. If you upload that, or you you do the data yourself, or however it will happen, and you find that relative volume is really important, well then, I would compare it to other things for gap ups that don't have the same opening range thing. Like, what about just a five minute candle low or, you know, all of these different entry criterias. And if them, with that relative volume filter applied, are all roughly the same, I think you've kind of disproven that it's your your filter. So just like someone would robustness test an indicator by trying a various inner indicator values, I think you can do a little bit of the same just by going through and making sure that your is it your thing that picked the entry?

Michael: 13:24

Is that the thing that matters? Or is it just you could have picked anything kind of random, and it's just the fact that it's a high gap with now a high relative volume. That's the thing that matters. And in which case, you've got what you mentioned as a universe filter. So now you have you you know that stocks that are gapping big with big relative volume are that important, and instead of worrying about optimizing the filter set, you're just left with another problem, where you have to just try to find what is the most ideal kind of entry condition that you need to see.

Michael: 13:56

And that's a little bit harder because it's it's harder just to to kind of put a a cruncher or to to to, you know, shrink that value down. But still, you know, you're playing in a universe of stocks that are more likely than not to fade a gap. And now, it's just to kind of figure out at what exact signal am I gonna end up buying at. Does that make sense? So instead of yeah, you're doing kind of the opposite approach of of what most people I think end up doing.

Dave: 14:26

Yeah. So it sounds like to me that what you just described there, you sort of theoretically, you found a filter that applied a lot of edge to the idea, and then you and that was with a given signal. And you said, okay. Well, this filter could be applied to other signals that are similar, that have a lot of overlap with the same universe, And you picked another signal that's, you know, sort of adjacent to the first one, but not exactly, and verified that that works with that signal too, so you have more confidence in the filter that you apply to the to the first signal. Is that is that fair?

Michael: 15:18

Yeah. Let's I think I think another example might might do it a little bit more cleanly. So, you know, let's say you're you're trend following on a long term chart. Right? And you're using just a very typical, I'm gonna buy a stock as it breaks out to a new twenty day high, and then that's my that's my mechanism.

Michael: 15:37

And then you get an equity curve, and that equity curve looks fine, just random kind of noise. Then you crunch that down and you find out, okay, well, you know, stocks that are experiencing strength with some other indicator, RSI or something, they they work much better on that on that breakout. Well, now you gotta figure out whether or not buying the twenty day high was the thing to do or whether or not buying stocks with a high RSI was the thing to do. So you could, again, complete the treat those independently and say, okay. Well, I'm gonna grab every stock with a higher high RSI, and I'm gonna test buying a breakout of a twenty day high, and I'm gonna buy a breakout of a momentum surge, or I'm gonna buy a breakout of a new all time high, or whatever.

Michael: 16:24

And if all of those strategies look similar, then you know the predictive thing was the RSI. The predictive thing was not the twenty day high break. So you can you can kind of pull these two things out, which doesn't mean anything bad. It means, well, now you know you're looking to buy stocks with a higher higher a high RSI. You just gotta figure out how to do that optimally.

Michael: 16:48

So that just becomes testing different different signals. You found the correct universe, you're just trying to pick the the best timing moment inside that Yeah. Universe.

Dave: 17:00

Yeah. And I think that's it's important to think about signals as events, alerts, and then Mhmm. Filters as a way to filter those events. I I really like the way Trade Ideas does this. There's there's alerts, and there's filters.

Dave: 17:19

I I think that's a great way to separate these things, and it's it's a good way to understand what's going on because events because they're they're two very different things. And, you know, alerts or signals, those are points in time like an event has occurred, and that's what you can use to generate the strategy from. And a filter can be applied to that alert and say, okay. When am I taking that trade, and when am I not taking that trade? And that boils it down to in a very crisp and clear way, like how to create a strategy, and I think that's very useful.

Dave: 18:04

One thing that a lot of people well, I've heard a couple things that people will say over the years that kinda make me cringe. One of them is, hey, I can make any basically any signal profitable by just applying risk management to it. I've heard people say that, man, I've I've heard some pretty well known people say this. Like, basically take a random entry, and if you apply a good risk management, then you can make it profitable. I mean, it's such BS, and but yet, I hear people say this pretty common, and it's so crazy.

Michael: 18:46

I instantly thought of the Jurassic Park reference. It's like, they spent so much time asking or if they could, they didn't ask if they should. It's like, could could you? Probably? Well, why?

Michael: 19:01

Like Yeah. You know, why why would you ever yeah. I guess, if you assume, like, you know, everything's somewhat randomly distributed and you had a a good risk reward on it, you could probably get there. I would say, probably gross of costs and and things like that. Yeah.

Michael: 19:18

But it's like, why? Yeah. It's not it's not why we're doing this. That's not

Dave: 19:23

But I think that's a good Yeah. It's a good exercise to think about the most watered down signal you could think of.

Michael: 19:31

Mhmm.

Dave: 19:31

It's like basically a random event, like literally random. And that's think about that situation. Literally a random event, a random signal, you know, you could create a strategy from it. You could probably do some filtering to make that work, but you're gonna really like, the filters are gonna be doing all the work there, and you're really gonna have to understand what the filters are doing a way more deep way than if started with a really strong signal that had a lot of edge to begin with. You kind of had your job is easier when you apply filters to something like that, because you know it's unusual, and you know it has edge right from the beginning.

Dave: 20:13

So I think it's a good way to think about signal strength and how important it is.

Michael: 20:20

Well, and I would say if you if you picked a random event, this actually might be in now that I think what it might be an interesting exercise to do. So if you picked a random a bet event, and your bet was the stock was gonna close higher at the end of day. You picked a random event, Like, I will literally random. You buy on a random you split the I'm starting to get interested. But you you split the day into one minute candles.

Michael: 20:48

You buy any random one minute candle on the day, and then your question is, does it close higher than that candle or not? And you put that into the cruncher or or again run the filter set however you wanna do it, and you end up coming up with a really predictive filter. Well, at that point, are you doing just what we described earlier, where you're identifying a universe in which that universe has a really high probability of closing positive. And although you'd never trade that, and you know, it'd be dumb to do, you've you've narrowed down a whole bunch of stocks that have a probability of a higher probability of trading lower, and then that could be the basis of a strategy. It's an interesting thought experiment.

Michael: 21:37

But yeah, just don't know why you wouldn't. Yeah. And how are you gonna get confidence in that? Like, so if you're the person, you know, going back to the guy who's saying, oh, I could take any signal and just apply risk management to it. It's like, yeah, until you're down six months in a row just because of some random BS, and then you decide your your whole experiment here was stupid.

Michael: 21:59

It's like, again, through large numbers, you probably could, but just because you can doesn't doesn't mean you should, doesn't mean it's a good idea. Yeah.

Dave: 22:09

I think I think it's a good way to think about strategies, and that how much work is the filter doing to create the edge. So there's some strategies where that the the initial signal is pretty weak, like, sorta like what you described, like, take pick a random candle, maybe not random, but, like, pick a candle early in the day to give it some time to play out. Right? Yep. You could come up with a filter set that makes it look good.

Dave: 22:42

So in that sense, the way I think about it is, hey, the filters are doing a lot of the work here. The filters are the important thing for this strategy because the signal's not is, you know, is less important. It's less it's less strength there, but the filters are doing most of the work. That's a perfectly valid approach, I think. And and and that there's strategies that exist that that do exactly that.

Dave: 23:14

In that case, you know, because the filters are more important, you're gonna have to think more about that. And any more rule you apply to it, you're gonna have to, you know, put it through your process and, you know, make sure it tells a coherent story. You know, all those things we talk about to make sure that strategy has edge now and going backwards, but also, does it tell a coherent enough story that I can believe that it's gonna be predictive going forward?

Michael: 23:46

And that's the that's the real important bit is, you know, there's there's two ways to get to it. It's the first to focus on signal, right, and then focus on signal in kind of a raw way, and and test whether or not that signal on its own has any edge. And if it does, then you apply filters to it. But, you know, what we're describing is kind of the complete opposite. You're you're testing almost the filters alone to see if they help narrow down a universe that has any kind of edge.

Michael: 24:14

And then you're testing the optimal filter on top of it. And neither's probably more correct than the other. Maybe that is, but at the end of the day, what you've ended up doing is you've ended up creating a signal that you've proven one way or another has some sort of predictive edge, and then you've add filters to it, that filter universe to things that that also have that predictive edge. So, you know, whatever the confidence, whatever makes you feel good about about placing the trade, when again, you're in a drawdown and and all the stuff that we talk about, that's fine. I would personally lean to the first one, where I wanna be sure that whatever I'm doing has that the signal itself has that edge, not because it's mathematically better or anything, but because it will help me feel better about the system, because I have identified that I think that that signal should do something for some reason.

Michael: 25:22

And that's what I think we gotta get back to, is that there was there was a reason that you sat down and said, I wanna test when a stock breaks a new all time high, and I wanna wanna test buying that. There was hopefully, it wasn't random. Hopefully, there was some reason that, you know, you saw a bunch of stocks break a new all time high and then rip, and and all whatever it ended up being. The that was your reason to get going with it. And then if that turns out not to do anything, then it becomes really hard to, you know, justify the rest of the work, where hopefully you have a whole list of things like that.

Michael: 26:01

So you test a new all time high break, and there's really nothing there, and it's really not that interesting, fine. Then you just move on to the next thing. Whereas if you do it the other way around, I think you would kinda keep yourself in in a loop, and unnecessarily loop a little longer.

Dave: 26:17

Yeah. I mean, this is where trading experience comes in. I mean, this is where discretionary trading and discretionary traders kinda have an edge in creating systematic strategies is because they're pretty familiar familiar with signals, and they already have some sense of whether it's gonna work or not. And the more experience you have, the more confidence you can have right away that there's something there because you've seen it, you've experienced it. And that's I've I've seen discretionary traders just have a really good sense of that, and they can save a whole lot of time in back testing and and creating systematic strategies because they know what to look for, and they know that there's some edge there already.

Dave: 27:08

So they end up saving a lot of time, you know, versus somebody who starts from a purely back testing and, you know, data mining approach, there it's gonna take longer. It's gonna take longer to to have confidence in something and to you may end up going down some rabbit holes following data when you don't have an intuitive sense of it. And this is why when I first start working with traders, I want them to I mean, ideally, they they continue doing what's working for them. I don't like to, like, rip out what they're doing if it's working for them and, you know, change that up. But they because they have something that's already profitable that that's working for them, it's a it becomes a really good starting point to have something that's tangential to that, that they're not trading now, like maybe it's just outside the rules they're trading now, and and start from the ground up with this, you know, an approach that's adjacent that they're already pretty familiar with.

Dave: 28:16

They're already familiar with the universe that they're trading, and starting from the ground up to make that automated because that's really pretty powerful when they can continue doing what's working for them, but also create something that's completely automated and ends up being completely additive to what they're already doing. Mhmm. And, you know, there's a reason that I don't suggest that they go and do something that's they're completely unfamiliar with. Like, it needs to be or the really what I'm seeing it work well is it's adjacent to what they're doing. It's not exactly, but it's adjacent, and they're familiar enough with the space that they can get confident in it, and they know there's some edge there just because they've experienced it.

Michael: 29:00

Yeah. And I I recommend the same thing in a way that you you don't want to again, like you mentioned, expose what happens too much, and you don't want to give up an edge when you have it. But that's kind of how it happened to me as well, where all of a sudden, I I switch from being discretionary to systematic when systematic started to beat discretionary. And I think that's a good way of doing it. And it might not happen with your first system, like, say you you're you're trading discretionarily, and it's working really well, and then you create your first system, and I don't know, say you're you're making $5 a month from your discretionary trade, and now you're making an extra thousand dollars a month from your systematic trading.

Michael: 29:48

Okay, that's a cool little bonus. And that gives you a really good motivation without, you know, if you had just changed entirely, and you're not taking a $4 hit a month, you've just added a thousand. And now, I remember that being a pretty big moment because I'm like, well, shit, I didn't do anything for that other Yeah. That other thousand. Right?

Michael: 30:09

It's it's way less money, but or but it's I didn't I didn't have to do anything. Once I came up with the idea, you know, I just had my process and everything was on rails. And then just your your attention kinda slowly shifts, where you're like, well, I get get either gotta size up that one or I gotta build just a couple more of them, and then all of a sudden, great. And then as that kind of shifts over, and then hopefully, you get to a point where you're like, well, why am I doing the discretionary side anymore? It's no longer, you know, the the primary breadwinner or whatever it is, and the you do the equation of the amount of time it takes versus the amount of, you know, returns you're getting from that time.

Michael: 30:50

And that ends up being a pretty interesting way to do it. But the other thing I wanted to mention is that you talked about how, you know, the the discretionary move becomes easier, and I think that's also something that's been that that I hold true as well. And we've mentioned it once or twice anyway, but it's it's why I still have, I call it a slush fund or a play account or whatever it is, but it's it's way smaller than my systematic one. But I take discretionary shots, and I force myself to do it. I have to take from the swing trader side, swing trading side of things, a trade a week.

Michael: 31:24

And there's sometimes on Friday where I haven't taken a trade, and I'm like, well, I have to. And I just look through and I just take take a shot, because it forces me to do just that, to see sometimes I have like a really wicked trade, and I go, okay, well, was that just b s luck, is there something there? And it it kind of forces me to go back and study that thing. So that's something I know you're probably gonna vehemently disagree with me on this one. But I do think that that kind of forcing yourself to do a couple discretionary things, even if it's super small.

Michael: 31:55

You know, say your your main account's, I don't know, a $100, and your and your play account's 10. And if it's just, you know, you're not gonna make or or lose a huge amount of money on a $10 account. But, you know, maybe if you do real good, you get like an extra couple thousand dollars at the end of the year, you waste on on something dumb from your discretionary ones. But it's more of a fueling, and and like we're talking about with the signals, it it just gives me more signals to test because and I saw something that I thought just from a pure technical point of view looked interesting, and I took the trade. And then it gives me that rabbit hole to go down because I saw the signal work.

Michael: 32:34

It's an n of one study, but I saw the signal work live in real time, and then it it does this good feedback mechanism where I say, well, was that does that happen often? Is there a trade that I can make because because that thing occurred?

Dave: 32:51

Yeah. I think that's a I think that's a good approach. It's not one I take, but it I do do kind of a similar thing. I don't actually take the trade, but I'm I am noticing, you you know, I put myself in a position to notice the unusual things, which I think is kind of the same thing. And instead of, like, taking a trade just to sort of, you know, for grins, I use that as like, okay.

Dave: 33:15

Let me use that as input for strategies that I could come up with or think about. So, yeah, I think that's a I think that's a really good way to do it, and I like that it's on a certain cadence. You have to you have to do it. So it's sort of, you know, you're forcing yourself to be thinking about this, and, you know, you don't have a great idea every week, I'm sure, but I'm sure there's some that have come out. You know, I'm sure there's strategies that have come out of that process because you've put yourself in a position to notice unusual things.

Dave: 33:44

I mean, that's really what this is about.

Michael: 33:47

Yeah, and and some of my best ones, and the main reason to do it is is the whole signal thing that we're talking about, is that I'm I'm, you know, as a CMT and as an analyst, and then someone who, you know, goes on CNBC sometimes and talks about what's going on in the world and and all of this, I'm always there looking at it anyway, because I enjoy it. Because it's, you know, other some people watch sports. Right? I I go through charts. And, yeah, I find it just it's putting myself in the right environment to, you know, see something that couldn't, you know and if having your own account is something that turns people off, I know Dave's monitoring anyway, but I'd say some skin in the game.

Michael: 34:35

Like, even if it's a paper trading account, even if it's one of these prop firm accounts that you can pay for, and you know, you give them $50 and they give you $20 worth of buying power, and and really, you don't have much intention of ever passing it, but you're just there to to hit some buttons. I do think it's a it's a good exercise because all of a sudden, the the signal is the thing that you noticed and you saw the trade happen live, and then you can do the whole process we talked about where, again, your n of one study or if you have made the same trade a bunch of times, you've you've got some data that it works, And you can then go through and start to look and see, okay, now let me test the signal raw, let me do the filtering behind it, let me and it just gives me a better understanding of how to how to push. Like, at what point if I see a a filter come in, did I recognize that on the trade that I had taken that worked? And just that little bit of extra confidence when you're like, oh, remember that trade, I made a couple $100 on it or something.

Michael: 35:36

Even though it's not a significant amount of money, it's still something that is it's real because it's a amount of money. Right? It's a dollar amount.

Dave: 35:48

So let let me bring this back to some things that I have said over the years, and and I've seen traders take that to the nth degree. And one of them is, you know, casting a wide net. So you find a signal and then, like like, so let's say opening range breakouts. You go and you find, you know, your back test trades every opening range breakout.

Michael: 36:18

Mhmm.

Dave: 36:19

So that that is an approach. And sometimes I'll get a trader that'll send me a backtest, and they'll say, Dave, what do you think about this backtest? And I'll see that there's 200,000 trades in it. Before I even know anything about the strategy, I can tell that there's gonna be a lot of noise that they're trying to trade in that signal. I don't care what it is, over what time period.

Dave: 36:51

If you're over that amount, I can just tell there's gonna be a lot of noise in there. So I like my starting points to be tighter than that. So, you know, I've got strategies that I start with sometimes 20,000 trades, even sometimes I mean, I've got one strategy. My starting point is, like, 400 trades. And the the the the hard part is starting with a tight a much tighter starting point is you have to understand that signal very well.

Dave: 37:30

Like, you have to know it very well because every decision you make, every filter you apply is gonna be the the trade set is small, like you're gonna it's even more important to think about and be careful with those filters that you apply. So that's one mistake I see traders making is starting with just a huge backtest that's I know is watered down right from the beginning. Now that That Go said ahead. I've got I've got another point after

Michael: 38:09

No. Was just gonna say, at at that point, you're not I would argue that's not even really a signal at that point. Right? Especially, you know, not to not to pick on the example, but every stock, not every stock, but I bet you if you did the math, it's gonna be 80 to 90% of stocks are gonna break their opening range at some point throughout Yeah. Throughout the day.

Michael: 38:30

Right. And so you have to look at it and you have to say, am I capturing an event, is I think the way you described it for for your signal, or am I just capturing something that's that's going to happen? The same way you could say, oh, well, you know, I'm gonna buy things that break the 20 high. Well, even the shittiest stocks on Earth, at some point, are gonna make a a new twenty day high, like, unless they're actually just stair stepping to zero forever through random noise. So I think before you can even quantify something as a signal, it has to be specific enough that it's not gonna happen to everything all the time.

Michael: 39:10

Like, it Yeah. It's it has to be a rare enough event that it's yeah. Otherwise, I would say it's not even a signal. You're just saying well, you're almost to the point where you're saying buy the stock on a random candle throughout the day.

Dave: 39:23

Yeah. I I think that's right. And so I think with opening range breakouts, you're right. There's gonna be a lot of stocks, just by definition, break in opening range. You're gonna have to combine that with a filter to create a universe where that actually makes sense.

Dave: 39:43

So but that does let me tell you my favorite kind of strategy to create. And that's where somebody looks at the signal, and they're like, yeah, there's nothing there. So basically, what I I call it, it's like hiding in plain sight. So to find stuff like that, it is they are gonna be pretty common. So basically, it boils down to, it's it's not really a science.

Dave: 40:21

It's more of an art. But I'll I'll I'll look for strategies where the signal is, like, what I call hiding in plain sight, and most traders would look at it and think, yeah, there's there's nothing there. But combined with a good filter set and a good column library, you can come up with good strategies exactly that. And I love that they're like, nobody would look at them and see and think, oh, yeah, there's edge there. Like, I I I've got several strategies like that, and I I love them sort of that.

Dave: 40:52

It's like a it's like it's in stealth mode, kind of.

Michael: 40:57

Yeah. And but I think probably what's happening there is you're doing you can at least explain them to yourselves, and I I think that's that's really the important bit at the end of the day, whether or not, you know, the initial question being is the signal predictive. The the thing I think you really need to ask yourself, because it's like we mentioned, you can kinda turn anything into a good equity curve, is that do you do you trust do you have some reason to believe that the signal is good? Because, right, the user was saying, well, how do I know what ones to chase down and what ones to leave alone? I would argue even more than whether or not it's a a pretty initial backtest.

Michael: 41:42

It's like, do I have a reason to believe that this is some mechanism, that this is some behavioral finance or or whatever is occurring in trading, that I'm like, oh, I I have either seen this work a lot and it makes sense to me why it works, or I could just kind of completely understand why this would work. And even if it doesn't test amazingly off the bat, if you have that like underlying belief and confidence about why the thing happens, then I think it's worthwhile pursuing a bit. And if you're just doing it because, you know, you saw it in a video or or something like that, then it's probably not worth that time. Those are the ones when I see someone else saying, oh, this is a trading style that I I use. Those are ones that I'll have, I'll test still, just in case they're right, but I won't spend a lot of time on.

Michael: 42:37

It's either gonna work right away or or I'll just I'll shelve it for later. But the ones that I see in the market, and I'm like, wow, that's really interesting. Then those are the ones that I'll be a little bit more dogmatic about exploring different filter sets, and hold times, and stop loss, because I have just that more of of the confidence. So like we talked about, you need the confidence to run the system, because it's just gonna go and do its own thing. You need the confidence to allocate your time to build the system, and that comes from some sort of belief to saying, yes, this is I I believe this should work, and I can kinda articulate the reasons at least to myself on why it should work as a market mechanic thing, as opposed to just, right, two lines crossing for the sake of two lines crossing.

Dave: 43:25

Yeah. I think that sums it up pretty well. But yeah, I think, just to go back to the original question, how strong does the signal need to be? I think the takeaway here is, and I feel like we've probably confused some people, but I'm think sure we do that

Michael: 43:43

a lot.

Dave: 43:44

Yeah. I think though that there's not one answer to this. You can have you can have profitable strategies where the signal is not super strong

Michael: 43:57

Mhmm.

Dave: 43:57

But you're gonna have filters that are quote unquote doing most of the work to for creating the edge. And that's gonna be a bit harder to come up with. They're gonna you're gonna have to think harder to avoid curve fitting, etcetera. But and you can have signals that are stronger, and I think that's you're gonna have an easier road to come up with a strategy like that, but they're probably harder to come by for new traders to to come up with that signal that is strong sort of right off the bat. So, yeah, I think there's it it there's not one right answer, but I think it's important to understand that continuum and how, you know, trade the the signals are probably easier to come by on one side of it, but they're gonna be harder to you're gonna be have to think harder about coming up with a strategy that works with them.

Dave: 44:52

That makes sense?

Michael: 44:53

Yeah. So, yeah. And and that's that's kind of what I was talking about too, with the the amount of time you want to allocate to it, I think is gonna be part of it as well. And and, you know, every now and then you're gonna stumble across something and it's gonna be like predictive and amazing off the bat, and you're gonna know that that's something that you really wanna dive into. But other than that, you're you're just kinda guessing at the end of the day where you're allocating your time.

Michael: 45:18

So make sure you're not wasting it because it's a like a thought you had in the shower randomly versus something that you've seen work in the market. Because although it won't, hopefully, it won't cost you any money because it hasn't gotten to the point where it becomes a real system until it's vigorously tested, it can cost you something that's like more important, which is just your time. Yeah. And I I think lastly, it's good to just point out, there's gonna be a lot of dead ends anyway. Like, yours has been I I don't could not I don't want to know the amount of time that I've spent thinking that something has an edge, and testing it, and coming back hours later and saying, no, there's nothing there.

Michael: 46:01

There's gonna be a certain amount of that anyway, so you just wanna make sure you limit that, I think, as much as you can, the best you can.

Dave: 46:08

Yeah. And also, that's not a unproductive end to the like that, understanding that is gonna help you in the future by avoiding that or, you know, there's always something you can learn from doing that that you can apply to other strategies in the future.

Michael: 46:25

Well, as always, you know, we we love the questions. They they create very interesting discussions. So make sure you're, you know, you're continuing to reach out, you're continuing to ask whatever whatever questions come to your mind, and they might make it on the show as well. And and thank you for the user and any follow ups. You know how to get a hold of Dave, so you you can ask that as well.

Michael: 46:47

But as always, I'm Michael Nauss.

Dave: 46:49

And I'm Dave Mabe. We'll talk to you next week on Line Your Own Pockets.