Line Your Own Pockets

Michael: 00:00

Okay, everyone. Welcome to another episode of Line Your Own Pockets. We're kinda following up a little bit of last week's episode where we talked about, you know, how to size up a a system or strategy that's doing so well. And we're gonna go over a email that we got from a listener. Again, we love the the feedback and the interaction.

Michael: 00:17

We're gonna talk a little bit about just position sizing as a whole. So, Dave, do you wanna talk about the email you got and the nitty gritty of the the question that was asked?

Dave: 00:29

Yeah. So this email was from Scott p, who I got permission to use his name on here. He's somebody that I've interacted with for some time. He's on my newsletter. When you get on my newsletter, at the end of the welcome sequence, there's an offer for a free call with me.

Dave: 00:45

So he, he signed up and had a free call. He's a really interesting guy, and I'd like to have him on the podcast or at least there's another whole area of of his trading that I think would be super interesting.

Michael: 01:01

Cool.

Dave: 01:02

This week, though, he asked me a very specific question about position sizing, and I thought it was a great follow-up to what we were talking about last week. So let me just read what he says. Dave, what are what are your views on position sizing and percentage of capital at risk per trade? How much is too much on highest conviction trades? Do you subscribe to the Kelly criterion, half Kelly, etcetera?

Dave: 01:27

Mhmm. So I thought it was a a a good chance to go over all the various position sizing approaches that I've used and and then talk about the one that I've looked at pretty recently that I I'm not using yet, but it's really I've got it in the back of my mind trying to figure out the best way to use it. And that's the Kelly criterion. It's probably the most interesting one.

Michael: 01:52

Yeah. So yeah. And we'll go when we get there. We'll go over a little bit for for those who haven't heard of it, just the math behind it and then all of that. But, yeah, when it comes to position sizing, just like we mentioned, with a couple things, this is one of those overlooked, I think, especially for newer traders, but hugely important.

Michael: 02:09

You know, a lot of new traders are show me the place on the chart in which I should buy. Right? That's the focus I think of the majority of people who've, you know, haven't been in the game for even more than couple years. Right? It it's all about the setup.

Michael: 02:24

Where do I buy? Where do I where do I sell? And it's one of those things just like risk management where good position sizing can make or break a strategy very, very easily. And, I think even more than, you know, quote, unquote, good or bad, uses of position size is inconsistent. Position size can really kill a strategy because just the way the fates work is if you're, you know, you're risking a certain amount on trade a, and then you're risking double or triple that on trade b.

Michael: 02:53

Trade b is definitely going to be a loser. It's just the way that the the trading gods work. So, I guess we can just go through some of the most common ones. I think the most common that a lot of people do, especially in the investing and and more of my space, the longer term trading and swing trading space. It's just dividing your account.

Michael: 03:09

Right? If you're planning on taking up to, you know, say, 20 or 10 positions at once, you'll say, okay. Well, I'm gonna take 1 tenth of my portfolio into each position. So then even if the risk is fairly wide stop loss, it reduces that. You know, I not generally the most optimal thing out there, but the beauty of that is simple, I guess, in the moment and simple to operate.

Michael: 03:32

You have a $100,000 account. You're gonna split your trade into 10 trades. You're taking $10,000 trade. So, I think you're probably gonna be much fancier than that, but that, I think, is a really good starting point for a lot of people.

Dave: 03:45

Yeah. I think for longer term investors, that's certainly the right approach. I mean, with my long term money, that's exactly what I do. Yeah. There's nothing wrong with that.

Dave: 03:55

The shorter your time frame, the shorter, the the shorter your trading is, I think you'll need to get fancier with your approaches. I think you're totally right. Mhmm. So the first one that I the the the one that I is first on my list here is just constant number of shares. Now this is I think for a lot of traders, this is maybe their default, but I think for retail traders specifically, this is a bit naive.

Dave: 04:31

It's really interesting to look at when when you're doing backtesting, looking at constant share size versus some other approach, you'll see a very wide variety of equity curves just by changing the position size. So it's really important to think about it. And, yeah, I used to really not like this approach, but, actually, recently, I've found some areas where that it actually makes sense. You see this a lot in the prop trading space where you're not, you know, you don't have a certain doll number of dollars in your account that you're trading. You have a certain level of buying power.

Dave: 05:16

And just by the mechanics of how that dynamic works, constant share size could make sense in some situations. So I used to knock this and think it was super naive, but I'm not so sure about that. I'm I use this approach actually in a couple of my models.

Michael: 05:33

Really? Okay. This could be interesting because I I will need a lot of convincing to get on board with you. That the way I look at it is that you take 1,000 shares of stock a or stock b knowing that stock a is 10 times more volatile than stock b. It it, it doesn't make sense to me.

Michael: 05:53

Same as even when it comes to looking at something like even for the same price. Right? So you're taking a 1,000 shares of of a $10 stock and $2.10 stocks. And one is a quantum computing, whatever, and the other is like general electric. Right?

Michael: 06:10

So I don't it'll be interesting to see. So try to square that circle a little bit for me because right now, as I stand, I could not disagree more.

Dave: 06:19

Yeah. So for this specific strategy where I use this, I'm basically using my entire account. So I run out of buying power almost every day with this strategy. So just for the simplicity's sake of using a a a constant share size, like, anything different than that is it gets a little bit more complicated when you're bumping up against buying power issues. So I am going back and looking at this and figuring out how to optimize my buying power with this strategy in a way that would use a different approach.

Dave: 07:05

But it's not, like I said, it's not the best strategy, position sizing strategy in the world, but it's not as bad as I once thought it was.

Michael: 07:16

The the only place and I've been just thinking while you're talking. The only place, like, I could see a rationale for using just the constant share size every time is if the securities you're you were trading was relatively constant. If someone comes to me and says, okay. Listen. I trade the SPY.

Michael: 07:31

And then I'm like, okay. Well, on most days, you're you're gonna be fine because it's gonna have roughly the same volatility from day to day. And then, you know, maybe a a COVID event or something crazy happens, and and you have to adjust a little bit there. That I could see. And and the reason for that is that that person, it doesn't matter how fancy they really get in their positions.

Michael: 07:51

The it's it's gonna be roughly the same. So that's the only thing. Or if the strategy was somehow normalized for the volatility of the name. If you're saying I'm going to trade stocks between, you know, that that that move roughly between x and y, say, ATR or something on the day, I could I could see that as a devil's advocate. But, again, I just think it it'll always be, suboptimal.

Michael: 08:12

Like, I I just can't think of a way that Yeah. Even for the spy guy, I would say you could probably get better if, you know, you you did something based off of intraday volatility or stop loss distance or or one of the other ways we'll probably talk about.

Dave: 08:27

Mhmm. Yeah. So it's, like I said, probably the worst of the approaches, but not as bad as I once thought it was. Let's just put it that way.

Michael: 08:37

Interesting. I'll be interested to see if if you can come up with a an argument that I'll that I'll accept for right now. It's just it's yeah. Yeah. For for me, I I would say, yeah, if you're trading the same instrument, maybe explore it that way.

Michael: 08:51

Dollar the the percentage of the portfolio or or or dollar amount, to me, that's already in in infinitely better way because at least there is some correlation between price of security and volatility, at least in some way. Right? So you're buying more of a a $1 stock share wise than you are a $100 stock. So there's some kind of normalization there. But, yeah, share size just you know, it seems weird.

Michael: 09:17

You're you're taking Berkshire Hathaway with a 1,000 shares or you're taking, I don't know, x y z, dollar loss of computing stock with a 1,000 shares.

Dave: 09:27

Yeah. With this type of strategy, yeah, you would have an upper limit on price for sure. I mean, that yeah. So that that's the that's the caveat here.

Michael: 09:36

Yeah. You have to be dealing somewhere in in a in a dollar range. Then I could see again, just like I talked about with, like, that there's people out there. I know it's weird to us, but there's people out there who sit down every day and just trade Nvidia, and they will only trade Nvidia all day. And and that's all it is.

Michael: 09:51

And again, for those people, I I could see the argumentation, but, I again, I still think it's it's a little bit suboptimal, especially in the in the world that we live in where you could you could come up with, on the fly metrics that I think would be better.

Dave: 10:06

Yeah. Yeah, I agree. Alright. Let's go to the next one. So and this is sort of related to your first one, is the way I look at it.

Dave: 10:14

It's just constant dollar amount. So instead of a constant number of shares, you got a constant dollar amount.

Michael: 10:20

Yeah.

Dave: 10:20

And I think of this in the same bucket as your, constant, percentage of your account just because it's the same thing.

Michael: 10:31

Yeah.

Dave: 10:33

I like there are strategies I use this for. I I still think this is kinda suboptimal, but there's some strategies where I think it makes sense as sort of the easiest thing to do. Like, even with your the the first thing you mentioned, even that's gonna be suboptimal because you're gonna have a a wide variety of ATRs for for names that you trade. So doesn't really like, you can tell it's gonna be suboptimal in some situations. So, yeah, what what do you think?

Dave: 11:10

Do you use this in any of your strategies?

Michael: 11:13

The the percentage of account, yes. The the dollar amount, no. But just kinda kinda like what you mentioned with your share size, there's some normalization going on. And I think a lot of the differences with some of the systems that I use, and this is probably what we'll get into a little later, is that there are no stop losses. Stop losses happen in some sort of trailing.

Michael: 11:35

And this is gonna be the difference between something like a trend following strategy versus something like a, you know, a day trading strategy where you generally always know you're out. I think when it gets to the point where you're doing something like a long term trailing stop or even I have a rotational strategy that just rotates every single month, there's no level to play off of that I could even be cuter with. So, again, for simplicity, I use that. Again, even with those, I think you get cuter with some sort of, right, ATR multiple or or or something like that. But, yeah, it's one of those things that when I tried to get fancier, the improvements were incredibly minor, but the complexity added to the strategy was major.

Michael: 12:19

And I I think that's something we'll talk about more of is that, especially if you're doing something in the moment, you have to balance those between I have the absolute perfect optimal position size versus, you know, I need something that I can hit the button relatively quickly. And, you know, if you if you're making an extra 1% a year because you're you're really pushing that lever, it's probably not worth it. Right? It's probably better just to to come up with a simpler system overall. So, yeah.

Michael: 12:48

So long story short, I I do use the constant dollar amount, which is percentage of my account, which then I I just adjust, like, monthly based off the the value of the account.

Dave: 13:00

So the other big thing that that is a difference here between constant shares and constant dollars in any sort of position sizing approach is when you go to optimize the strategy, you're gonna get very different results Yeah. Based on the position sizing. So with a constant share size, all the expensive names are gonna be overwhelm all the cheap names. So you you're gonna get a very different optimized result when you do it that way versus another approach. So, definitely, it's so fundamental to a strategy that it's just it's it's it's fundamental, and you have to think about that first before you do any other optimization or so or Well, and

Michael: 13:48

and I would even say it's part of, like, robustness testing. Whereas if if you take your account, you say, okay. I'm gonna take the top 10 symbols that are outputted from this strategy based off some sorting mechanism, and the strategy works great. And then you take like, I'm just I'm still I'm just ignoring your constant share size because I still can't get that in in my head. But you do something like an ATR based or volatility based and the system completely falls apart, then I think you've kind of even done some robustness testing there where you're like, why, right, in a constant split, but then not in a, an ATR kind of or or volatility adjusted adjusted system.

Michael: 14:29

So, yeah, I think, yeah, it's absolutely paramount. And in in in that way where, yeah, if it doesn't work, you know, if there's a massive change in differences and small differences to your position sizing, then, yeah, you you gotta look at it's just one of those, you know, scratch your head a little bit and think think why for a while.

Dave: 14:48

Mhmm. Yeah. For sure. Alright. Ready to go for the next one?

Michael: 14:52

Sure.

Dave: 14:54

The next one is similar to the other, but I I I see these as getting more and more advanced each time. So Yeah. This one is just simply sizing by ATR. Yep. So, you know, total total amount, you know, you would take fewer shares of a higher price or a higher ATR stock, more, more shares of a lower ATR stock.

Dave: 15:21

And I and I like this because it really is normalized. I mean, this is how when you go to optimize strategies and you do constant shares versus ATR, just when you do ATR, it just feels like things come into focus much better.

Michael: 15:40

Yep.

Dave: 15:40

Like, the stuff the the really important features in the model start bubbling up, and that it it it really feels like things come into focus.

Michael: 15:51

Yeah. Because you're taking the luck of and and I think that's really what it is. It's just random luck if you're doing, say, constant share size and and say you're only trading $10 stocks. Really, when you're testing that, your strategy comes down to, was I right in the incredibly volatile $10 stock, or was I right in the this $10 stock that didn't move? And that's gonna vary your your back test result insanely.

Michael: 16:19

It's gonna be huge. Yeah. When you're doing something like vault normalized volatility, it's more about was I correct based off the setup, and then was I correct in that mode as opposed to anything else. So this is something that I would say most people should try to get to even if they're not trying to get super crazy advanced with their strategies. That could be a fairly simple math equation.

Michael: 16:42

You take the ATR and then and then you just say, okay. I'm gonna divide this by the amount of money I have or whatever it is and then go through and just come up with a pretty simple math equation that says the the crazier the stock, the less I'm gonna take, and and then go from there. I think that's a pretty easy one to do. And I think that's if we're talking about, like, the amount of complexity, the amount of bang for your buck as we kinda go up this ladder of complexity, I think share size to, the the constant position, I think that's a really easy one that I think generally will will do better in most scenarios. And then I think this is another leap again where you're not going too crazy with it, but at the same time, you're you're you're doing a a massive benefit for not a lot of work, I think, is is the easiest way

Dave: 17:28

to put it. Yeah. I think that's true. The other thing you that that you noticed you know, we mentioned last time where or just a second ago where when you're optimizing constant shares versus constant dollars, you get a a big difference. Similar thing when you do constant dollars versus ATR.

Dave: 17:47

When you do constant dollars, you tend to see high t high ATR names having a difference in the stock. So Mhmm. Or in in the strategy. So you get when you do when when you optimize, you start seeing ATR being important, things related to ATR being important.

Michael: 18:05

Yeah.

Dave: 18:06

So when you normalize it across ATR, that then, again, comes more into focus, more into actually what's, what affects the strategy and what's predictive in the strategy.

Michael: 18:19

And would you in the ATR bucket, and I think we're just to make sure we're clear for the audience and and for me, we're putting all volatility metrics. Right? Whether it's standard deviation, whether you're just doing, because I'm one of the day trading strategies I'm working on. It's just basically because it doesn't have a stop loss. It's basically just sizing it based off yesterday's bar and kind of clump all of those in, I guess.

Michael: 18:42

So when we're saying when we're saying this, we're basically just saying something that looks at the recent volatility of the stock itself and then sizes based off of of that because, of course, we won't know what the future volatility of the stock is or we'd all be trillionaires, but we're using that as a best guess estimate to say, okay, this name is crazy. So I wanna trade a little bit less. This name doesn't really move, so I wanna I wanna trade it a little bit a little bit bigger.

Dave: 19:08

Yeah. And I and, yeah, as you were talking earlier, I realized, yeah, volatility is probably a better way to put it. I just like using HR because it's it's easily available in a lot of platforms

Michael: 19:20

Most. Yeah.

Dave: 19:21

Without doing anything. Yeah. Like, basically, for doing nothing. So it's just very convenient to use it that way. It's kind of sort of a standard measure.

Dave: 19:29

A lot of people know it. So Yeah. I kinda use that interchangeably with volatility. Are you ready for the next one?

Michael: 19:37

Okay. Well, I so the way you took a deep breath before you said that means we're we're probably getting into getting into the complicated land now.

Dave: 19:46

Yeah. It's funny because this next approach is the way I made 100% of my trades for probably the first 6 or 8 years of my trading career.

Michael: 20:01

Can I can I try to guess? Yeah. R multiple. Some sort of r multiple trading.

Dave: 20:07

Yeah. Basically, r multiple. And Yeah. So and I don't actually it's probably not best to call it r multiple. What the way I like to think about this is you're sizing your position based on the setup.

Dave: 20:22

And I think that's super important because everything we've talked about till now has basically ignored the stop loss and the distance to the stop loss. Mhmm. And I think it's a fundamental thing for people to understand the the r multiple and expectancy concept. And so just just to review here, when I say size based on the setup, I mean, take a position size such that if your stop loss is hit, you lose a specific amount of money. Yep.

Dave: 20:55

So that that's what traders who use this approach say, that's what's at risk.

Michael: 21:01

Mhmm.

Dave: 21:01

The amount your the amount of the amount of dollars you have at work in the position is not what you have at risk. Really? Yeah. Really, what you have is is the distance between your entry price and your stock price. So once you kind of start to grok that, you realize that, hey.

Dave: 21:20

That's true. And, really, it it forces you into this mindset of okay. What do I have at risk here? What is really at risk, and how can I maximize my profit based on my risk? And that's an important concept that, and I would say it's advanced, but I didn't think it was advanced because that's the only way I traded for, like, 6 or 8 years.

Michael: 21:46

When it comes to, intraday trading, again, I I could not agree more. Right? Van Tharp, anything that Van Tharp did is great. He is kind of the, I don't know, pioneer or popularized this idea of of constant risk amount. The caveats that I'll give to that is, again, a little bit different for longer term trading because we cannot be insured of our stop loss the way day traders can as much.

Michael: 22:15

Right? The the amount of slippage we need to count for would be more with gaps and and things like that. Right? If I wanna stop it at $10 and it gaps down to $8. Right?

Michael: 22:26

You know? Well,

Dave: 22:30

I would well, I I don't understand why you wouldn't be able to use our multiples for that. You got a stop loss. Right? And if it gets passed, then, okay, you you lose 2 r maybe or 3 r or more.

Michael: 22:43

Well, and that's what the kind of the second thing I was gonna talk about is because in most cases, I found, at least for me anyway, you're, you're doing a lot based off ATR and you're you're sizing. So in if your stop loss is 1 ATR away and let's say that's what you're using for a stop loss, now we're talking little tiny differences between sizing based off the stop loss and then just doing an ATR multiple trade. And that's kind of where I've gotten to where if I test the difference between a constant share size and a percentage of account, the the amount's different. It's huge between those two tests. If I test the difference between the percentage of the account and ATR, the distance is way smaller because I'm already normalizing volatility in some way.

Michael: 23:30

And then if I do the ATR to the stop loss, it's derived on ATR anyway. It's it's a much smaller leap. So could be some benefit there. I definitely think for intraday trading, probably way more, because you can do some math, like, how much money am I look willing to lose in the course of a day. And, you know, you can size those things.

Michael: 23:51

I would just say probably a little bit less impactful, but still impactful for for the kind of longer term ideas because a lot of what you're doing is based on ATR in a lot of cases anyway.

Dave: 24:03

Yeah. So I I guess in that situation, I I think I'm talking about something a little different here. So when you're when you size your we were talking about sizing by ATR, and you're coming up with a stop loss based on ATR. Yeah. I look at that as different than a setup.

Dave: 24:21

I I think of a setup to be independent of what the ATR is. So you're looking at a range. You're gonna get in above the range, and you're gonna get out below that range. That range is I I consider that's part of the setup. Yeah.

Dave: 24:36

So size based on the setup regardless of what ATR is. In fact, if you, you know, if it's a high ATR name and you got a really tight stop, that's that could be really good because you you'd be able to get a lot of, share size, relative to the ATR in a in a situation where you're you've you've minimized your risk there. So, yeah,

Michael: 25:01

I look at it as different as just simply sizing by ATR. I'd say the difference is, yeah, you're so you're you're basically saying my stop loss goes based off the chart regardless of ATR, and then I'm going to to take a trade from there. What I'm saying is a lot of the things that I do, especially mean revision of trades. Right? So you're looking for things that are tanking.

Michael: 25:22

Well, you're trying to buy near the low. So the only way or the way that I use currently to figure out where the stop loss is gonna be is I take the ATR and I apply some distance to my fill price to the ATR, and that's for the stop loss. So I'm saying in those kind of cases, sizing based off ATR or sizing based off stop loss, you're essentially it's the same it's kind of the same horse. Right? Because the Yeah.

Dave: 25:46

Yeah. I see.

Michael: 25:47

The the the it's not the setup is buy it here, risk another 2 ATRs from where you bought it from. So in those types of scenarios and same thing when it comes to a lot of trend following strategies in reverse, I'm buying it and I'm not using any sort of chart indication of where to get out. I'm buying it and I'm putting a 2 ATR trailing stop on it. So, again, in a lot of these cases, the stop loss if the if the stop loss is the same as the the sizing metric, then again, little improvements maybe, but probably a lot of the same thing. But like what you're talking about, I'm gonna buy it here.

Michael: 26:21

I'm gonna put my stop at the low of the day. Right now, you're talking a completely different different thing. Right?

Dave: 26:26

Yeah. Yeah. So I think it's important to think about that. You can also do, you know, calculate our multiples and expectancy across all these if you want to if as long as you have a stop loss. And I think that's valuable to do.

Dave: 26:42

So, yeah, I think that

Michael: 26:43

I think that goes to the last episode. So if you haven't watched it, certainly go back and do it. But we were talking about how to size up a strategy that's working or size down a strategy that's not working. And our multiples can be great for that because if you just tie that back to the percentage of your account you say I am risking 1% of the value of my account that's my r value as I'm doing better my account's increasing in size then that r will go up if I'm doing poorly and it's decreasing size that r will go down so that's one of the major major benefits of that same with percentage of account size because that percentage number is going to gonna grow or shrink so you kind of tying the 2 together where you're saying your stop loss but also by using that, you're kinda building in a longer term plan for sizing up positions, sizing down positions, that kind of thing.

Dave: 27:29

Yeah. Alright. You ready to go to the next one?

Michael: 27:34

Sure. Man, I

Dave: 27:35

Alright. This one

Michael: 27:37

Well, we gotta be getting close to Kelly because I don't think we can get much more advanced without hitting Kelly.

Dave: 27:41

Yeah. The next is Kelly. And it really it it I think it's the most interesting one.

Michael: 27:47

I think we missed one. It's just the martingale. Right? That's isn't that the way you size things? You just if you're wrong, you double up next time and Yeah.

Michael: 27:54

Right? Have a plan to run away from your wife to Cuba if it doesn't work out.

Dave: 28:00

Alright. So let's get to the last part of Scott's question. Do you subscribe to the Kelly or half Kelly criteria, etcetera? I've known about this for a long time, and I didn't really look into it forever. But recently, I've read a book called fortune's formula, which is about the Kelly criterion.

Dave: 28:22

It came out, 2006 or so.

Michael: 28:26

I always look these up the second you, you say so I don't need to. Here we go.

Dave: 28:31

This is, actually a very entertaining book. You would think that something written about a math formula would be super boring, but this one's actually a quite an entertaining book. So we'll put it in the show notes.

Michael: 28:43

Cool. It does have an audiobook, which is good because I'm very Good. Very audio. I don't read. I listen.

Michael: 28:50

So, good.

Dave: 28:53

So the real interesting thing about the Kelly criterion, and you learn about it as you read the book, is it's literally proven in a mathematical way that this is the optimal way to do position sizing for a strategy for growing the account. There's no other way that can be better than the Kelly criterion. Now the problem is when you go to look at the formula and you go and apply it to your account, you're gonna have some enormous volatility in your equity curve. Yep. So that's the hard part of doing it.

Dave: 29:39

But but to me, the it's very intriguing to know that, okay, with just a couple data points from your strategy, you can come up with literally the optimal way to trade it for for growing your account.

Michael: 29:53

So let's let's break it down here. I'm not gonna read the whole math equation. Right? Because a lot of people listen to this, but I have Investopedia up. Shout out.

Michael: 30:01

Right? Probably the greatest resource for this kind of stuff. Yeah. So Kelly basically uses just two data points, and it is the historical win rate of the trading system and then the traders win loss ratio. So you're taking your win rate and your win loss ratio, and then you're doing some fanciness.

Michael: 30:18

So the which really, those are the massive the only two data points that kind of everything else in when you're evaluating it is derived from. Right? How much you make when you're right, how much you win when may or lose when you're wrong, and then, what's your probability of being percent? What Kelly ignores, and I think what you were talking about this as well. See, I'm pulling up my old CMT knowledge.

Michael: 30:40

We had to study a lot of this. Kelly doesn't study path. That's part of the problem. So we talk about path dependency, Things like our multiple trading and and the percentage of the value of the account do kind of focus on path dependency. So path dependency being in what order do the trades come in.

Michael: 30:59

Right? If if you're right in a big block of trades something like our multiple trading will grow your account really really fast and then and then vice versa really really small. Kelly doesn't adapt that much because it takes the total win loss percentage and the total risk reward and then goes from there. It will change, but just a little slower, especially if you have a lot of data to give you that historical win loss rate. And that can cause a lot of issues when it comes to those massive swings that you're talking about.

Michael: 31:28

Right? So, you know, say you have a a 100000 trades, and you're risking 1% of the value of your account. And the last 100 trades were all losers. Your account value, the amount you're risking per account, that's gonna start to correct pretty quickly. But Kelly, since it's gonna be based off those full 100,000 trades unless you do something that I think people call rolling Kelly, where they only take the last little bit, Kelly is still gonna look at your long term, win loss ratios and everything, and it's gonna give you roughly the same the same idea.

Michael: 32:04

Now it is based off the percentage of your capital, so there will be some adjustment there. But it's just it's just slower and slower like we talked about with sizing up and down your position sizing can be good, can be bad. Right? Sometimes it's good not to react when something happens very quickly to your equity curve, and Kelly will be slower in that. But, yeah, it's it's a lot.

Michael: 32:25

And a good exercise just to plug it into your current current model and see what the swings would be because it can get pretty wild.

Dave: 32:33

Yeah. I I'm not sure that I mean, there's nothing in the Kelly criterion that would prevent you from adjusting your account size with every trade. So if one goes bad, you would adjust it just like you're doing with, basic, you know, percentage of your account.

Michael: 32:50

Yeah. But the so what the rolling Kelly is meant for is not so much the percentage. It's the other side of the equation. Right? The historical win rate.

Michael: 32:57

So say you go way back and, you know, during the, you know, eighties nineties or whatever, your win rate was 70%. But recently, you've started to get that edge erosion, and now your win rate's around 50%. But if you have enough data, if you have enough trades happening a long time ago, that win rate won't adjust quickly. You know what I mean? Because you have such a data set to go off of.

Michael: 33:21

So again, some people will use what they call rolling rolling Kelly. And and so for day traders, it might be, I'll take my win rate over the last month or or week or something. I won't include my entire win rate, because what if the strategy was doing really well 2 years ago and then it really sucks now? It might be better to adjust the the formula more frequently. You know what I mean?

Michael: 33:43

Because depending Yeah. Depending because it's all based off how much data you have. If you if you have 2 trades, right, Kelly's gonna change really quickly. If you have 200,000 trades, then Kelly's gonna change a little bit slower.

Dave: 33:56

Yeah. So I've I would think about it differently. My assumption was when I've studied this is I wouldn't be basing it on my actual results. I would base it on the back test and, you know, be conservative in the back test to know, okay, what what what is my real win rate here likely to be, and then what is my, you know, profit and loss likely to be, the the two things that entered into the equation. I think that's where people have trouble understanding this is the they get those numbers wrong because there's some tail risk in their strategy that they don't realize.

Dave: 34:47

So they're not using a correct number to do the sizing. They're doing something that's too big. Mhmm. So I think that, I I I haven't used this in any of my strategies, but I know that I've been too conservative over the years with sizing. So it's I'm this is in the back of my mind as I come across new ideas and new strategies, and I'm trying to figure out the best way to apply it.

Dave: 35:17

I know, full Kelly is really, you can go do the math, and it's really yeah. It's optimal, but it's gonna be pretty brutal.

Michael: 35:26

I just want to say going from saying I'm undersizing a little bit to going full Kelly is just like saying, I've never done drugs before. I'm gonna hop in it like heroin. Right? Like Yeah. There's a couple steps that I think you should take before there.

Dave: 35:40

Yeah. So, yeah, it's definitely has me thinking, and, like I said, I haven't applied it directly to any of my strategies, but it definitely has me on the lookout for ways to apply it or something similar to it, but because of that because of that book and because of thinking about it more.

Michael: 36:01

Yeah. And and using half Kelly or a quarter Kelly is very common. One of the things that I was reading for the CMT essentially talked about how there's a lot of funds out there that will run their back test based off of, a more standard position sizing and and then kind of feel it out and then run it on full Kelly and then start to adjust the percentage of Kelly down so they're comfortable with the the drawdown. Right? So so maybe full Kelly, it's just you know, you have a drawdown of 47%.

Michael: 36:36

You're like, listen. I can only deal with a drawdown of a third of that. Right? Then you run it again with with that number and and see if that's a little bit more, more tolerable, I guess, to you you personally. But just like anything, it's gonna have to do with if I'm 20 and I'm getting into systems trading and and I got no house, no kid, no whatever, and I'm making a bunch of money, the amount of of volatility that I can withstand is gonna be gonna be way different.

Michael: 37:01

So you're gonna have to have to adjust that, but usually, very rarely, I would say it's it's full, Kelly, because I think, you know, you get some 80, 90% drawdowns. And even though it's it's one of those mathematically, it might be the correct thing to do, but then you actually have to envision yourself sitting there and and stomaching it. So I think I think of volatility or or some amount of drawdown based Kelly or volatility based Kelly would make sense. And then also, again, look at that rolling. At what point are you going to update, that system?

Michael: 37:34

Right? You know, say you're running you're running a back test, charting the win rate, I think, is actually and probably the win loss ratio is probably a pretty good exercise there. Because if all of the win rate or the win loss ratio happens at a certain part in the equity curve because you're using the entire data, that could, you know, give you a little bit of pause. But, yeah, it's something I've been playing with and trying to get a little bit deeper into myself because I like the main thing I like about the Kelly criterion is that it we when we talk about everything systematizing, it it systematizes your position size kinda forever. Right?

Michael: 38:12

As long as you're okay with the drawdowns that this will occur and you've you've done a percentage of Kelly correctly, well, now you're never really thinking about position sizing again. Right? It's just what does the Kelly criterion say and and, you know, re inputting those numbers based off the back test and then the win rate, the win loss ratio, and all that kind of stuff.

Dave: 38:33

Yeah. And the other thing it does is when you when you when you do it that way, you realize just how important it is to get those numbers accurate. Yeah. Come up with the accurate win rate, not just what you've seen over your trading in the last bit. Like, what is the theoretical and best number you can come up with?

Dave: 38:49

And the only way you're gonna be able to do that is through back testing, understanding, you know, what your back test might be missing, some theoretical things about, you know, some tail risk in your strategy that you might not be aware of or something that hasn't happened yet that could. Those are important things to think about. And, yeah, they should it's like a it's like a nudge to I always feel like traders should be always you should have a little trouble sleeping. They should always have something in the back of your mind thinking, okay. What what have I missed here?

Dave: 39:32

What am I missing? It these numbers look good, but what could happen? What am I missing here? How can I gather some more data to get more confident in what I'm doing? But, yeah, it's good to it's good to lose a little bit of sleep every time just, to to to wonder what you're missing.

Michael: 39:55

Yeah. And I I think, you know, kinda what you alluded to there too, it's very important to to track the differences between the back test and the real world. Right? Because it would be very unfortunate to size based off of Kelly, based off of a back test, and not have done that work to see, like you mentioned, what what is the real win rate? Because it's only using win rate and risk reward.

Michael: 40:17

So if any of those numbers have any significant skew in them to the back test at all, you are risking either way too little or way too much based off of and probably way too much based off of what this is. And it's already a pretty significant, risk when it comes to if you're using full Kelly or even kinda half Kelly. But, one thing that it's mentioned in a lot, I was trying to remember on on other than the CMT where I heard of this a lot and and the Market Wizards. I remember coming up over and over again in that Market Wizards book series is that different these guys used Kelly in some way. Because to get into the Market Wizards, you had to take some pretty wild risks.

Michael: 40:56

So, you know, to do that, that's that's what the position sizing like. So, the other thing that I really like about it is that it adapts to, very low win rate strategies and very high win rate strategies, which I think is one of the things that it is kind of key to it because, you know, risking 1% of your account, which is, you know, standard, like, just our multiple trading, that can that works as well. All these work to some degree. It's almost it's just which one's optimal. But I really, really like it because you could put in a win rate of 20% in your strategy, and it's gonna it's gonna adapt to that really, really quickly where, you know, when it comes to losing 8 in every 10 trades, that equity curve is gonna look way different depending on whether you're using Kelly or you're using some sort of our multiple type kind of trading.

Michael: 41:47

So really high win rate strategies and really low win rate strategies that it adapts adapts very well too. So I know a lot of people who do, like, revision to the meme. Like like, they're the kind of guys who are trying to catch the top and the bottom. And when they do, they just kill it, but they only do it maybe 30% of the time. A lot of those guys will use Kelly because it keeps them when they're wrong 7 out of 10 times from just getting blown out.

Dave: 42:13

Yeah. Yeah. It's a good point. It it adapts. Yeah.

Dave: 42:16

I like it.

Michael: 42:17

So is there any other, I guess, weird ones that we haven't thought of? Like, I'm I'm just trying to go through, and it it's funny because it really all of it boils down to, you know, you're you're picking some metric, and you're saying I'm risking some percentage of the value of my account based off that metric, whether it's, right, ATR or even the the equal sizing of your account. You're basically saying, okay. 10% of the value of my account in each of these trades, it's still in a way saying I'm risking some percentage. But I just wonder if there's something wacky out there that no one's ever thought of that that we could kinda discover.

Dave: 42:53

Yeah. I'm sure there is. This is all I was able to think of when I have been brainstorming. But, yeah, I'm sure there's others out there that, make might make sense. Yeah.

Dave: 43:04

If anybody's listening and is has some that we haven't thought of, yeah, send me a note. Let me know.

Michael: 43:10

I think there's one thing we can agree on, and it's definitely don't try Martingale. I can't even imagine, Dave. I was reading this in the CMT textbook about to get tested on it. Like, why are you even putting this out there in the ether to people who are going to be managing large sums of money? Why would you even teach them?

Michael: 43:30

It's just it is a position. So any of those who don't know, Martingale basically just means you bet, I don't know, 1% and if you lose, you bet 2%. And then if you lose again, you bet 4%. And then eventually, you win and you make all your money back and some. And, you know, it doesn't take a math genius to know.

Michael: 43:46

You're basically setting yourself up for an eventual explosion. It just matters what it is. So, yeah, they're teaching this, and I'm reading this. I'm like, you guys shouldn't even tell people that this is a thing unless the whole chapter was don't use Martingale, and this is why. Yeah.

Michael: 44:00

I it just seemed seemed crazy to me.

Dave: 44:03

Yeah. That does seem crazy.

Michael: 44:05

Alright. Well, if anyone thinks of anything, again, put it in the comment. If there's any cool sizes, definitely do that. This is another one that was given to us from an email. So we we love reading emails and messages, and, I've got a couple.

Michael: 44:20

I don't know. I think I tagged you on Twitter that we're gonna maybe chat about in a bit. So, again, we love the community feedback. We love all the, everything that we've gotten when it comes to, you know, the positive feedback and all that. It'll love it because, you know, we just kinda jumped into this and said, hey.

Michael: 44:37

Let's let's see what's happening. And to be honest, it's been blowing my expectations out of the water for the the feedback we've been getting.

Dave: 44:44

Yeah. It's been great. And, yeah, I particularly like the I think this episode's gonna be one where it doesn't get a lot of views, but it's gonna be very valuable to the people that do view it. So At the end of the day a few like that. Yeah.

Dave: 45:01

I think there's gonna be a few like that that are underrated, and I suspect this will be one of them.

Michael: 45:05

We'll just have to clickbait the next one with, with something crazy just to to make it up for it. But, it's been great. It's been a great discussion. I think we got a lot of a lot of things that people can go and and rework their strategies and and test it. But until next time, I'm Michael Knotts.

Dave: 45:23

And I'm Dave May. Hope you join us next week on line your own pockets.