Line Your Own Pockets

Michael: 00:00

Alright, everyone. Welcome back to another episode of Line Your Own Pockets. We were talking a little bit in the in the backroom there about how we're we might trigger some people with this one because we're gonna talk about curve fitting, which is I always get told is the devil. Right? It's the worst thing you can do.

Michael: 00:15

As a systematic trader, there's a lot of people out there. I hear that anything that you do that's got a back test has some amount of curve fitting in it. So I think me and Dave will probably be aligned together on this, but I don't think we'll be aligned with a lot of the people out there. So it's gonna be fun to upset some people.

Dave: 00:30

Yeah. So it's a huge topic. It's one that I have obviously thought about for twenty years. And it it struck me as I was coaching this one particular trader that I remember very, very distinctly. And we were looking at his strategy, and it was painfully obvious to me that he could improve his strategy very easily by you know, basically, it was, an early version of the cruncher.

Dave: 00:56

We got rid of 20% of the trades. Mhmm. And it was so obvious to me. And yet this guy said, oh, man, you know, this looks good, Dave, but, man, I'm I'm really worried about curve fitting. I don't wanna so, basically, he was I could tell he was terrified of curve fitting.

Dave: 01:12

He was completely stuck, not wanting to change his strategy. You know, his strategy has been working for a while, but here but I'm showing him in black and white. Here's how you can improve it. And yet he was unwilling to make any change at all for this fear of this boogeyman.

Michael: 01:26

Well, let's, let's take a step back and first just do some definitions. Right? So, you know, curve fitting, essentially, we're talking about the statistical phenomenon of I forget who it was. I said, you know, you torture data long enough. You can make it say whatever you want.

Michael: 01:39

And the idea is that it is very possible, and it is something that you'll see a lot where you you have an equity curve that just goes straight up into the right. And what's happening is, generally speaking, you're making the data fit. Right? That's the best way I think of of putting the curve fitting is you're just saying, I just wanna get rid of everything, every negative trade that's ever happened in the the history of this. I just wanna get rid of that, and and I want to just have it look like it's a perfect 100% win rate and and all of this.

Michael: 02:10

And, of course, that's an extreme example, but there are abilities to to kinda do that. And and some of it, I I think we'll talk about, you know, what causes curve fitting and and, you know, ways around it. But I always I think the example I remember from statistics class, they said you wanted to find out, like, the average height of a person in your school, and you stand outside the the basketball court, and you start to you only survey those people, and you say, yeah. Everyone at my school is, like, six five. Right?

Michael: 02:35

So, you know, there's there are ways that you can you can kinda torture data. But I think and we haven't talked about it before, but I think me and Dave are gonna kind of agree that very easy to avoid, and we'll probably give you some techniques to do that throughout the podcast. But when we say curve fitting, that's what we mean. It's just torturing the data until the equity curve looks great. And then you take it to the real world and and you just get you get killed as soon as it starts going live.

Dave: 02:58

Yeah. So that's the so I've given a talk at S and B Capital about this exact topic, and I'm gonna go over the same sort of things I went over there. The first thing I like to think about here is what's the worst that can happen? Now I I've actually pulled everybody on my mailing list a few months ago and said, okay. Tell me an a real life situation where you've curve fitted and what happened?

Dave: 03:24

Like, what was the worst thing that happened to you from curve fitting? Because I you know, early on in my trading career, there was some situations where, oh, I was like, okay. Yeah. That's probably curve fitting. And, you know, I put something live.

Dave: 03:37

And with quote unquote curve fitted, nothing really happened. Like I see these horror stories sometimes where like people go live with something that's curve fitted and they claim their trading careers over. They lost a whole bunch of money and like, okay, I don't really know how that would possibly happen. Like, you have to be kind of you'd have to be so naive and doing stuff so wrong for that to happen that it just doesn't make sense to me. And and I didn't I got a couple responses to this email, but there were no horror stories.

Dave: 04:09

Like, nothing awful is gonna happen if you accidentally curf it. And we'll go into more detail about that.

Michael: 04:15

But Right.

Dave: 04:15

That's the one thing I like to say. Okay. What's the worst that can happen? And it's really not terrible.

Michael: 04:20

No. I I think in most cases, if you take a equity curve that's curve fitted and then you apply it live, it shouldn't be a the next day you lose everything. It should just be potentially that was the top or near the top of the equity curve, and and you start to to kinda tick down from there. Anytime anyone says, okay. I tried this, then my career was over.

Michael: 04:41

That to me is just a risk, like, position sizing thing. That's not even your strategy at all. If if you're doing that with a strategy that you have just taken live, then you're gonna do that with anything else. Right? You're just gonna blow up anyway.

Michael: 04:54

Right? It could be a great strategy in the world. You're probably just trading it. You're probably just trading it too large. I think it's it's one of those things.

Michael: 05:01

It's it is a legitimate fear, but I think it just get kinda blown out of proportion. Right? It's one of those people are are worried that they're gonna take an equity curve that looks amazing. And just like you talked about, the next day, their career will be over. And in most cases, it will just be, okay.

Michael: 05:16

When you start implementing this thing live, it's probably not gonna perform that well. And I think that kinda leads to the first thing I want to talk about is on how to deal with it is that you're before you have curve fit it, hopefully, you've gotten a whole bunch of data, and it turns out that whatever your trading style or flossy is without any optimization at all has some edge. And then the worst case scenario is you're just gonna go back to that that base strategy that has edge. So maybe your initial strategy, you know, has a profit factor of, like, 1.2. And you curve fit it all the way up to it has a profit factor of five.

Michael: 05:50

You would imagine the worst case scenario is you start trading it, and it sits somewhere closer to 1.2, and it's still an okay strategy. It's just not awful. I think the problem could be is if you start with a a strategy with a profit factor of, like, point five, like, an objectively awful strategy, and then you curve it that all the way up, and then it comes back down to a a more losing strategy after that.

Dave: 06:10

I like yeah. I like the way I like that. I I think it's a good way to think about it. So here's so what I did from our presentation with SMB is said, okay, let's say that you were trying to curve fit. Let's say that was your goal for whatever reason.

Dave: 06:27

Let's say that was your goal. How would you go about curve fitting? And I've got five ways here that you can use to maximize curve fitting. Now, of course, you never wanna do this. Mhmm.

Dave: 06:37

But let's let's just go through this exercise. And then the idea is let's just do the opposite of all these, and you can

Michael: 06:45

Yeah. Sounds fair.

Dave: 06:46

So alright. How to curve fit? Step number one, start with a small number of trades in your back test.

Michael: 06:53

Yep.

Dave: 06:54

That's probably the biggest thing because if you if you just don't do that, you're gonna have a hard time really curve fitting. Like, it's gonna be harder than you think to actually curve fit.

Michael: 07:04

And I I see that all the time, I know you don't follow these people as much as I do. But the ones who have, like, you know, this bearish indicator has flashed in the market, and, right, the whole market's gonna collapse. And then you look at it, and it's happened, like, four times in the past. And they're like, yeah. Market's definitely the other four times this happened, the market was doomed, and you're just like, oh god.

Michael: 07:25

Right? And that's, you know, that's no different from normal statistics, right, to go back to, you know, trying to find the average height of a person. If you grab four random people, maybe, you know, you're you're right, or maybe you just found four real short guys or four real tall guys or one really, really short dude, and it just skews your whole statistics. So it's yeah. That's that's one that you can carry forward from normal statistics as well is that sample size is gonna be, in most cases, king.

Michael: 07:51

Right?

Dave: 07:51

Yeah. And, you know, there've been people that I've talked to over the years where I'm like, you know, it's obvious that they're worried about curve fitting more than me, and then and we sort of, like, disagree about it. And then when we get down to it, they're trading a single futures contract or their their their system is just such that there are just so few trades where they have to relax their rules so much that the signal just disappears to get enough trades in your backtest. Like, it's just it's almost a completely different world. And if you start I honestly think that if you if you trade stocks and you have large number of trades in your backtest, you're almost immune from this.

Dave: 08:31

Like, you're not completely immune, but you it's just not something that is in the should be in the forefront of your mind all the time. It's not something that you you should be keeping, you know, keeping awake at night, worrying about curve fitting.

Michael: 08:44

Well, so let's define it a bit. Right? What is a large number of trades? You know, someone's, say, back testing a day trading strategy. Right?

Michael: 08:52

And the you know, how many trades would you say is an area? Right? There's obviously not gonna be an exact answer, but just somewhere that a listener could say, okay. That's that's good. Right?

Michael: 09:02

Are you talking tens or hundreds or thousands or thousands or or hundreds of thousands? Like, what where would you start to go, okay. This is you know, it's gonna be a little of a Loki's wager thing, but where is that kinda cut off for you?

Dave: 09:13

Yeah. It it's hard to say. Like, you it is hard to say. There's not a specific number. But but if I've got a backtest using Ameren Broker where I'm testing over a large period and I've got less than, say, 500 trades over several years, that's where I would you know, my antenna would go up and I would be I wouldn't, like, not trade the strategy, but I would be thinking, like, it would be I would be more worried about it in that case.

Michael: 09:38

Skeptical. Yeah.

Dave: 09:39

I would be skeptical. And I knew and we'll get we'll go into more of these rules because I think it I think this is think this is, like, a really good way to think about it. So you ready to go to number two?

Michael: 09:51

Absolutely.

Dave: 09:52

Alright. Number two, don't look at the equity curve when you're optimizing your strategies. That's a good way to curve fit.

Michael: 10:00

Mhmm. Yeah. And people just look at that title sheet of of winners versus losers, and they're not focused on, I guess, distribution of winners. Just an absolute number.

Dave: 10:10

Yeah. Just an absolute number. Even, you know, metrics that are really good like CalMar, you're still you're still gonna be losing the the really great story that that equity curve is telling you. And it's it's very easy to gloss over that and just look at a metric that looks good, but then when you look at the equity curve, you're like, well, okay. All that's from, like, through all the profit here is from, like, two or three days over this whole period.

Dave: 10:37

And, yeah, the the metric looks good, but one glance at the equity curve and you can see, okay. Well, that doesn't make any sense. Like, I would never trade that.

Michael: 10:44

Well and in most cases, probably what you've done, because, right, I think we've all hit this at some point, is you've probably really kind of curve fitted a day in the market. Right? So say you have a mean reversion strategy. You notice that you've tested five years and all of the profits in those five years. Well, you probably just found days at the market.

Michael: 11:02

You've curve fitted to this is what happened to the market when it bottomed and just ripped higher. And if you line up those days that you made a lot of money and you line up with what happened in the market that day, you're like, okay. I'm not I it was just leverage, what do they call it, leverage beta instead of alpha. Right? I basically just bought the market a lot and ended up working.

Michael: 11:22

So, yeah, I think that's a huge one. The distribution of them. If you have something that, you know, say it trades twice a day every day for five years and pretty consistently throughout that and it ends up working, then, yeah, you you probably haven't curve fit anything. That that would be the statistical chance of the fact that it just works all of the time like that would just be crazy to me.

Dave: 11:43

Yeah. Alright. So number three. And remember, this is how to curve fit. Right?

Michael: 11:48

Mhmm.

Dave: 11:49

Just follow the data. Don't think about it too much. Just look at your optimizations. Don't don't try to think too much about it. Just follow the data.

Dave: 11:59

Right?

Michael: 11:59

This is this is the one that I really hoped was in this. Like because, again, I I go in blind. You don't you don't share these with me, but I really hoped that this was in it because I think this is if you want, like, the holy grail of not curve fitting, it's probably this one. Right? All all of them are important, but it's probably this one where, you know, I I always just use an example of a trader who came in and said, oh, you know, I've got this great trading strategy.

Michael: 12:22

And it was like a five minute opening range break scalp. Like, was in and out in, like, a couple minutes. And he had, like, quarterly dividend growth or something in the in the backtest. And I'm like, so you that's a example of dude just followed. He put that in there because it it made it good.

Michael: 12:38

So, yeah, you always got to every time you add a filter or change a filter or do something, you gotta take a step back and and take a deep breath and rock back in your chair and think, does this make any sense? Is this is this something that has any should it in my mind have any kind of predictive value for what I'm trying to trade? And a lot of it can be that time frame mismatch. Like, you know, if you're in and out in a day and, you know, you're seeing how close you are to the two hundred day moving average pops up, it's probably not relevant. Right?

Michael: 13:08

Because you're in and out in in minutes sometimes. So, yeah, try to say, okay. Before I do this, does this make any sense? Right? Because you're probably not finding something that's like a a, you know, a weird correlation that makes sense.

Michael: 13:20

You're probably just again, that's a that's a big one for curfing. So I'm glad this one is in here. This was gonna be, like, my ad if it wasn't if it wasn't in there.

Dave: 13:27

Yeah. So the way I like to say it is, does this rule tell a coherent story with the signal you're trying to with the signal in your strategy? And if it doesn't tell a coherent story, then that should be like, you should be very skeptical about that, just like you said. So I I I love your example. So, yeah, I think, you know, the better you are at this, the the the more you know, the better your strategies are gonna be.

Dave: 13:53

Mhmm. So I I think this is a really important and and I'm glad you pointed out that, hey. This is kind of the holy grail. I think that's right. Like, you how likely is this rule to continue into the future?

Dave: 14:06

That's really what you're thinking. And that's the the the other so I remember working with this one particular trader. I was really emphasizing this point. You know, he needs to tell a coherent story. I sort of said like, I could tell that he was just following the data with these strategies he was coming up with.

Dave: 14:21

And I said, okay. Go through each one of these rules that you've applied to this strategy and tell give me the why. Tell me convince me why this rule should be in the strategy. And he came back, and it was like he was, like, trying to pass a test, like, just giving some answer that, like, a a fifth grader would say made sense. Mhmm.

Dave: 14:44

Just you could tell that, you know, it's the real person you're trying to convince here is not me or you. The real person you're trying to convince is your future self when you're in a drawdown and you're trying to decide whether to continue trading That's the person you're trying to convince. So think about think hard about that. When you're in a drawdown into the future with this strategy, the only way you're going be able to continue trading through it is if you really believe in it and you really believe in these rules that you apply. So that that it's it's just so important to to convince that future self that's trading through a drawdown or trying to decide whether to trade to a draw through a drawdown.

Dave: 15:28

I mean, every trader that ever quit did so doing a drawdown. Like, they're not doing it at all the time times. That's the hardest part of trading.

Michael: 15:35

And it it's funny. That reminds me of of when I used to trade prop in university, and it was the same thing. The office manager would just come out and randomly tap people on the shoulder and say, why are you in that trade? And if you stuttered, if you couldn't explain it in, like, very succinct rules, and this was all discretionary trading. But if you couldn't do it very quickly, it was get out of the trade.

Michael: 15:58

You're done for the day. And it was very much an exercise of you have to have a reason for everything. It's the same with adding a filter. Right? Why are you adding this filter?

Michael: 16:07

It's like, oh, well, you know, hey. It makes sense because it it, you know, it's it's high volume, and I'm looking for breakouts or very simple things, which, you know, again, if you've gotta squint real hard to right? This old, like, technical analysis thing. If you gotta squint real hard to see the pattern, it's probably not there. And and and same with this.

Michael: 16:24

If you gotta really think about why this makes sense, it probably doesn't. Right? It should be the ones that kind of, stand off the page. And that's actually you know, it's funny. We just got out of a long weekend.

Michael: 16:34

I spent some of the long weekend back testing and using the cruncher, and it was just getting to the point where the top 10 results just don't make any sense to me. So I I'm in the mode of I have to, you know, change something on my end or or rerun the report or or do something because even though, right, the use the cruncher, and I think it's a it's a great tool, it's just not giving me results that I'm comfortable adding to the system. So, right, you you have to go from there. And I think that that's gonna be very important for a lot of traders because kind of the whole point of this, you're trying to filter out the junk. Right?

Michael: 17:07

And, you know, you should filter out the junk trades, but you should also be filtering out the junk filters at the same time. Having that, I'm not in a rush to put the system live. I I want it to be something that I'm comfortable with. So when the drawdown comes, I can come in every morning and hit the button to to turn the system live even when it's, you know, taken money away from me.

Dave: 17:27

Yeah. So, usually, when you get to that point in the cruncher, you're you're either one or two things. One is you're probably done optimizing. It's probably ready. Or, yeah, you need to go back to the drawing board and figure out some other filter to add to your library that that will be more predictive than the ones you have in there.

Michael: 17:47

Yeah. And it it's it's just very much a you know, the graphs look good, but as I'm reading through the filters, I'm like, I just don't I just don't like them. And it's unfortunate that this is the answer because it's certainly not the answer, especially new traders are gonna like because it just means there's a there is a certain amount of, I don't wanna say, like, intuition or gut feel or any of those kind of nebulous things, but just experience of, you know, what makes a market move. So what I kinda suggest to the new traders is just to go back to the the reason. Like, what is the actual strategy trying to accomplish?

Michael: 18:23

And then just look, does that service that goal, or is it just something kinda completely different?

Dave: 18:28

Yeah. Yeah. I love it. Alright. Let's go to number four.

Dave: 18:31

Remember, how how to curve fit? Reason number four. Add lots of rules. The more, the better.

Michael: 18:36

Yep.

Dave: 18:37

Right? So every rule you add is a little bit more risk. It's a little bit more, you know, you're you're stepping a little bit more toward curve fitting, and you're increasing the complexity of your strategy. So if you wanna curve fit, add a whole bunch of rules.

Michael: 18:54

Yeah. I think that's that's another great one, and and it's one for one. You know, the best strategies you run is where the base looks amazing. You just come up with a good idea. You you ran it raw without any optimization.

Michael: 19:04

It looks great, and then you find, like, one or two things that just really make it better. And they're they're coherent with what the the story is of the setup and all of that, and then you're just good. It's like, you know, a lot of traders will say the best trades just kinda work for you right away. Whether or not that's true, I I don't know. The best strategies I found just work right away.

Michael: 19:24

Right? I'm not sitting there and and pounding my head on the table how to make the strategy great. It was pretty obvious from the get go. And, yeah, I'm I'm just thinking back to my trade ideas days kinda servicing clients. They'd give you back tests, and there was an icon for every Yeah.

Michael: 19:38

Filter they put in across the backtest. Sometimes you get, like, five or six lines long, which means they put in, like, you know, 40 or 50 filters into this, and you're just like, yeah. It goes back to the you know, by then, you probably don't have enough data or the following of the data and not thinking about it. It goes back to that as well. If you've added that many filters, you've probably, at some point, lost the plot.

Michael: 20:00

You're you're not you're following the data at that point instead of going, you know, deep into it.

Dave: 20:04

Yeah. Absolutely. You're you're totally right. Yeah. I remember, yeah, I've seen several strategies where when you go to multiple lines for those icons and trade ideas

Michael: 20:12

Mhmm.

Dave: 20:12

For the for the filters, like, wow. Okay. That's a bunch. And five lines. You're talking about five lines.

Dave: 20:17

That's that's amazing. I I don't think I've ever seen that many filters applied.

Michael: 20:21

Yeah. There's been some there's been there's been some wild ones where you used to where they probably used every filter that's inside

Dave: 20:28

Yeah.

Michael: 20:28

Trade ideas for for whatever pops up. And and, you know, the equity curve looks great, but it's just one of those you it becomes harder at that point, especially if if you're that far gone, then you have to really kinda hurt someone's feelings and say, there's no way this there's no way this thing's gonna I know you've to put that many filters in, you've probably been working a long time on this thing. And to just, you know, say, hey, this is this isn't gonna work. You know?

Dave: 20:50

Yeah. I have to do that with, traders that I coach. Like, you're basically bursting their bubble. And in fact, one of the traders I'm coaching now said, yeah, Dave, thanks for bursting my bubble. Like, it was basically, you know, he had his hopes up for the strategy.

Dave: 21:07

He thought he thought he had this really good idea, and it's like, well, here's why it's not gonna work. So the early you you you burst your bubble in the process, the better because it's only gonna get more painful as you get deeper into the process and and go live. And

Michael: 21:23

Well, I remember I built one that I thought and I think we we've probably all done this. You you thought you figured it out. Like, this equity curve was just insane. Yeah. And, yeah, very quickly, I I learned otherwise.

Michael: 21:35

But I'm glad I learned it then when my account was significantly smaller than it is now. And and, you know, it's it's always, like, it's good when you hear people about, you know, they've opened up, a $5,000 prop account and blew that up, and you're like, oh, good. Right? You're not doing it when it's hundreds of thousands of dollars or, you know, you're not blowing up 6 figure accounts. It sucks at the time, but it's good that, you know, you did early because you probably learned way more doing that than than later.

Michael: 22:03

So, you know, and another good example of why with a new strategy, you don't wanna go full tilt right away. Right? You wanna start that a little bit small with smaller risk and and kinda work up from there because if you've stumbled upon one of these things, you'll probably know pretty early.

Dave: 22:19

Yeah. There's a good chance you'll know early for sure. And that brings me to number five in this. Number five, if you wanna curve fit, don't follow-up after you go live. Just sort of assume everything's good after you get done with the back test and feel like your work is done then.

Dave: 22:34

Right?

Michael: 22:34

Yeah. And that is probably the part I always say that, you know, of the traders I've talked to, I don't think I've met one that was losing a bunch of money who had a a trading journal. I just don't think I've I've run into that. Most people, when they are losing a bunch of money, they don't have that. Because you're right.

Michael: 22:51

You could you know, in backtest, everything could be looking great. But if you're not doing that reconciliation back, which is way easier than it's ever been with, you know, Claude and things like that, you could just dump the raw data from both of them and say, here, are are these things working out as intended? But, yeah, not having that, I think, is is a huge deal where you're not going back and looking at the trades, and you're not going back and doing that kind of that reconciliation. So you really don't know if it's working or not until you look at the PNL at the end of the maybe the month at that point. And by then, who knows what's happened?

Dave: 23:22

Yeah. Anytime you add rule or make some change to your strategy like this, there is some risk. And and in that reconciliation process, you should be skeptical about it and going back and verifying and looking extra close at the change you made to see, okay, is this really working out like I expected to? And the better you are at doing that, the quicker you're gonna find out what's wrong. And if you're not looking closely, you can go a long time without noticing.

Dave: 23:54

And Mhmm. And you can lose money over a period of time. But but as long as you have a good practice about how you're trading and how you're reconciling, then you're somewhat immune to it. But but that's all the stuff that we've talked about, you should already be doing anyway. Like, this isn't, like, only reacting to curve fitting.

Dave: 24:14

This is just good trading in general. So I think I think that keeps good traders who have good practices from having to worry about curve fitting too much. It's sort of built into their process to keep it at bay.

Michael: 24:28

Yeah. And and especially the the going back and checking, I think, is is gonna save most people. Because then even if you're doing every other sin that we talked about, at least you're gonna you're gonna lose small, and you're gonna go, okay. Even if you, you know, you just fast forwarded to this part of the pod and you just listened to the to this area of it. And it's like, okay.

Michael: 24:49

Well, at least if, you know, you screw up, you're gonna screw up small. And, hopefully, you're gonna figure out pretty quickly what what ended up happening. If you're not doing those checks back, then you could be doing everything else right. And it could just be, like, an honest, like, mistake. Like, it could be, you know, fills aren't representing properly or or something, like, really simple.

Michael: 25:08

It still happens to all of us. And if you're not going back and checking it, then then you don't know. So, yeah, this is kind of the you could have all your ducks in a row. And if you're not doing this last step, it it doesn't matter because you're still gonna fall into a trap every now and then where, you know, you lag or or liquidity or or whatever it is, and you're you're not achieving that goal even though, again, you've you've done every part of this perfectly up until now.

Dave: 25:32

Yeah. Yeah. So so when you let me go through a specific scenario. So let's say you're adding a you've added a a filter to your strategy.

Michael: 25:41

Mhmm.

Dave: 25:42

You put it live, and now you're reconciling. One thing to do in that case is also keep track of the ones that you've removed from the strategy and look at those and make sure you know, compare those to the ones you're taking and make sure that the ones you're removing are consistent with the rule you've applied, you know, consistent with what you were trying to do and what you're trying to accomplish with the strategy. A lot of people will just look at those trades that you've made. I mean, there's and if you've removed some, let's look at those too. Compare those.

Dave: 26:16

That that's always a good thing to to do.

Michael: 26:19

Yeah. And it's why, you know, having the I think the base strategy is like a separate file is is so good. Not only to, you know, strip it down and reoptimize later, but, hey. You could look at it and you you run. So what I do is I run the base, and then I run the optimized version, then I run my own trades.

Michael: 26:34

And I'm looking between those. And, you know, every now and then, sometimes you'll get this, like, huge spike in the base where it was outperforming the optimization for some period of time, And that really makes you think. It's like, okay. Did I remove the right trades? And sometimes, still, yes.

Michael: 26:49

It was just a random you know, this one trade looked out, looked amazing, but still over the long run, it's it's better to remove it. Sometimes you go, okay. Maybe not. Maybe I shouldn't have taken those trades out, and then you can go back to the base and try to reoptimize with different filters or or whatever. Yeah, it's it's as someone who's not, like, a very detail oriented person, it is the the least fun part of it, but it's like you have to do it.

Michael: 27:13

It's like tracking my workouts. Like, I like just to go to the gym and throw stuff around, but I'm not gonna get better if I don't have some understanding of and I'm I'm definitely not as as diligent on this as a lot of people. But if I don't have some understanding of, you know, what was the weight and I used on this last time. I have no way to progress if I'm not at least recording the bare minimum that that I can do.

Dave: 27:34

Yeah. Totally. Alright. So let let's review these five, and then we'll I'll, I have a couple things to to finish up with. Alright.

Dave: 27:42

How to curve fit? One, start with a small number of trades. Mhmm. Two, don't pay attention to the equity curve. Three, just follow the data.

Dave: 27:50

Don't think about it too much. Four, add lots and lots of rules. The more, the better. Five, don't follow-up after you go live. So those are so, basically, just do the opposite of each one of those, and you really don't have to worry about curve fitting that much.

Michael: 28:04

It's funny. Like, as you're as you're saying those, I'm thinking of myself as a newer trader, and then I'm also thinking of just, like, so many traders that I've talked to that just feel like every single one of those. You know, like because one can lead to the other. Right? Just following the data can lead to low dataset, can lead to the but yeah.

Michael: 28:23

So splitting those out, I think, is is huge.

Dave: 28:26

Okay. And now so two two more things I wanna go over here. One is a lot of people think that a strategy is either curve fitted or not curve fitted when that is absolutely the wrong way to think about it. In fact, every strategy has a little bit of curve fitting in it. Like, we're predicting the future here is essentially what we're doing.

Dave: 28:46

So if you can think about it as not a binary yes, no, but more on a range, that can get you unstuck. Like, it's not the end of the world if there's a little bit curve fitting in there. Every strategy has a little bit of curve fitting. So it shouldn't be something that's like this boogeyman that you're you're completely scared of.

Michael: 29:03

Yeah. And and if you just look at it and saying, you're not trade if you're not trading every stock in the market every day, then you've chosen some filter to to go through. Right? The traditional one is, right, people only trade stocks in play that day. Right?

Michael: 29:16

Something that's gapping or doing volume or whatever. Okay. You've already done it. So, yeah, once you know that it it is a spectrum and you're just trying to stay away from the real egregious end of it, then I think it should make it a lot easier for for a lot of people to say, okay. Well, I'm doing it anyway.

Michael: 29:33

I'm just I'm just hoping that I I'm doing it the right way.

Dave: 29:36

Alright. One last, point that I wanna make here. So I I thought, you know, I think a lot about this because, you know, so many traders throw this term around all the time and are scared of it. Why do traders worry so much about fitting? And I have a theory about this.

Dave: 29:54

The theory is think back about when you learned what curve fitting was. You probably thought whoever told you, you probably thought, man, that is the smartest thing I've ever heard. Like, it's this thing that's you know, nobody sees. It's underneath the surface in the shadows, but you but this guy that you just told you about it knows about it. Like, he's in the know.

Dave: 30:15

So it makes you sound smart when you bring it up. So I think for that reason, it's brought up too much. People like, you're always a skeptic is always gonna sound smart. Right? So you're gonna bring it up more than it needs to be because you're gonna wanna sound smart.

Dave: 30:31

So I think it is brought up way too much precisely because of that, or I I think it's a big part of it. What was your take on that?

Michael: 30:38

I totally agree. Because it's funny from my world, from the the longer term world, there's always just a joke that, you know, bears bears sound smart and bulls make money.

Dave: 30:47

Yeah.

Michael: 30:49

Tom Lee, I think, is a a perfect example. Not not the drummer, the the investor, but he always goes on. He's always, like, bullish as hell. Right? And and wrong as much as probably everyone else's, but he makes a lot of money over time because he's always just expecting things to go up over time, which quite often they they do.

Michael: 31:06

And then you'll get the next guy on air, and he's giving you this big long thesis about why the world's coming to an end and all of this. And he sounds really, really smart when it happens. And then the guy comes on and goes, oh, we go up 75% of the time. Right? It it doesn't sound very good, but he's right.

Michael: 31:23

He's like, we go up 75% of the times. Yeah. I think it's the same thing. Of it might be like a social media thing too where we know negative things get way more traction than than positive things and and way more, you know, Internet points and and all of that. So Yeah.

Michael: 31:38

Yeah. I think that's part of it too is it's very it's very hard to build a really, you know, nice looking strategy with a good looking equity curve and, you know, post about that and start a discussion. And it's very easy to go on the comment and go, that's curve fitted as hell. Right? You suck.

Michael: 31:54

Right? That that took the guy absolutely no effort, but, you know, he's using a statistical term. Therefore, he probably sounds smarter than you. Yeah. Even though it's it took absolutely no effort to say, you know, that this thing's bad.

Michael: 32:06

Don't do it.

Dave: 32:07

I've seen that so many times. I remember one particular time seeing somebody somebody was posting about their strategy on on Twitter, and they said, yeah, made this change to it. And there was somebody in the comments that said, that sounds like carfeiting to me. It was like, oh my gosh. Alright.

Dave: 32:21

The carfeiting police have arrived. Everybody's safe. Right? Like, it was it's just some there's something about it that sounds smart, and you see it. It just bothers me to see that because there's just so much more to it and so much more to think about, and it's just it it bothers me.

Michael: 32:36

It's if you ever go on the Reddit, I forget what it is, like, quant trading or or something like that, That is every thread. Like, I spent, I think, like, an hour on that. I'm like, I I can never come back. Because if someone posts something, and then there's two people, and one is this is curve fit. It's awful.

Michael: 32:53

It's never gonna work. And the other one's like, why not just trade it and find out? And it's like, those are the two only responses to everything, and one of them makes perfect sense. Just trade it and find out. And the other one's just some guy you know, I think there's you could call it jealousy or or something.

Michael: 33:09

I've met the guy who says, trade it and find out. Has probably got some strategies that are making him a couple bucks. And the guy who says, no. There's no way it's gonna work. It's curve fit.

Michael: 33:17

It's awful. I bet you he's doesn't. And you just see it all the time in in social media and then just, like, humans at large. There's just a lot of negativity out there, I think.

Dave: 33:28

Yeah. And I think the person that says that is sort of missing the whole point, which is that the person that's you know, the person that's putting themselves out there saying that, hey. This is the strategy. I've made this change to it. They're not the only way they're gonna be successful is to convince themself that they're confident in the strategy.

Dave: 33:48

They're they're on a path to confidence for the strategy, and you need to be able to trade through the drawdowns. So there's no person to convince other than themselves. So it's it's just such a disservice to insert this, you know, silly comment about curve fitting for no good reason. Like, it's just counterproductive in so many ways.

Michael: 34:08

Yeah. But yeah. This is like I said, it's easy. Right? It's very easy to kinda, you know, tear something down.

Michael: 34:14

Someone paints a a beautiful picture and puts it up online and say, what do you think? And it's just much easier to say your picture sucks as opposed to actually painting your own or or doing your own kind of things. It's just kind of a symptom, I think, of the Internet. I I think part of it is, like, I have been doing, like, public facing stuff for so long. I'm immune to it, at least to some respects.

Michael: 34:36

But, especially people who are just going out there and want, like, legitimate feedback and, like, actual careful measured response. And it's like, no. This is Twitter. You're not gonna get you're not gonna get those those here. But it goes back to what we talked about with the the trusted kind of trading community.

Michael: 34:51

Right? You should find a group of people that you could send that to, and they could they know, they'll take some time and take a look at and try to give some sort of feedback or criticism or something that's valuable as opposed to just saying, no. It's curve fitted. You know, you suck. Stop kind of thing.

Dave: 35:05

Yeah. I love that. Alright. Any other things to talk about if we covered all the races here?

Michael: 35:11

No. I I will be interested to see the comments because I I think maybe we've laid it out our the argument good enough that we won't trigger so many people, but I imagine there's gonna be some. But it it's it's one of those discussions that nuance and and caveat, I think, really help out in saying, hey. You know, is this you know, what do you think about curve fitting? You can't just say it happens or it doesn't.

Michael: 35:35

Right? You can say, you know, our our initial and our kind of final response is that's probably a little bit overblown. But then you gotta explain, I'm looking at the timer, the the forty minutes of stuff that we've talked about until then to say, you know, what to what to avoid. But as always, you know, comments and and all of that, we we read those and and appreciate those. And I'm Michael Noss.

Dave: 35:56

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