127: Building Mean Reversion trading strategies with Cesar Alvarez - Part 1

Now, those of you who know Caesar's work would be aware that he is a mean reversion specialist. He has a wealth of knowledge on mean reversion trading. He's going to share a lot of that with us over this special two-part mean reversion series. So I'm really excited to be sharing it with you. Now in this first episode, Caesar is going to be discussing the high level steps we need to take to build a mean reversion trading strategy.

How to select a trading universe, and how to find mean reversion setups using simple but highly effective techniques. He also shares with us how to identify better quality trades by combining indicators. And he he gives us some really interesting insights into why strategies with a smooth equity curve may not actually be the best strategies to trade in the future.

Also, if you have any mean reversion questions you'd like to ask Caesar, then I'll tell you how you can submit questions for him, which we'll answer in the second episode in this series. So listen out for that. At the end I'll tell you more about that afterwards. Anyway, let's jump straight into this first part of the two part mean reversion series with Cesar Alvarez. Hope you enjoy it. Hi Cesar, welcome back. Thanks for coming on the show again.

And that is mean reversion. So I think a lot of people are going to find this one interesting today. Now, um, before we get started, how about can you just give us a quick little bit of background on yourself for people who uh perhaps didn't catch the other episode?

as a quantitative trader. I left about four years ago uh to go out on my own and, you know, do trading on my own and to write my blog. I just really wanted to write a blog and, you know, uh give back to other people and write about various research I like to do.

And my most of my trading is based on stocks and ETFs. And a lot of it for the longest time has been mean reversion trading. I do do other types of um strategies like trend following, uh breakout strategies, but I'd say definitely my my bread and butter and you know, where where I really like to do research is on the mean reversion trading front.

I'm not a you know, I'm not a day trader. I've tried it before and that's I just that's definitely not me. And you know, waiting, holding, you know, a month or two months, that that kind of time period tends to be a little bit on the longer side. that um you know also is not really for me. So I I like the I guess, you know, we're all trying to get something out of the market and that

the action level at mean reversion trading is kind of like the right level of action. And also I also like it'cause it it tests your uh fortitude. Definitely that's one thing that mean reversion does and um something that uh you know, we'll hopefully talk a little bit more about that a little later.

five percent of the time, I'd say at least, or don't even look at the symbols. Um, because if you look at these charts, you're gonna be going, Why the hell do I wanna even buy this stock? It's it's clearly in a downtrend, it's not gonna bounce and

Often those are actually the best ones. You know, so that's you know one of the hardest parts about mean inversion trading. Um one of the things I really do like about mean inversion trading is it's got a pretty good win ratio, usually in the mid 60s, so 60, 65% is often.

where your um winning percentage is which quite nice. And you can get some nice winners really quickly. So, you know, like I said, recently I had this last week or so, I had like a ten percent winner in like uh one day. This thing just bounced like crazy.

So you know that's one of the nice parts about mean reversion trading. Uh the other thing I like about it is, you know, these are quick holds. You know pretty quickly within three to five days whether you're in a good trade or not. Um so you know, those are all the things I really like about mean reversion trading.

small cap stocks, whatever. Um of course the main part of your strategy is, you know, how are you gonna measure mean reversion? You know, how you gonna measure you know, the stock is sold off. And you know, that's you know, that's the core part of the mean version. So we'll definitely spend a lot of time there. And then, you know, what other filters are you looking for to add to your um to your strategy?

So those are that. And then you know, you know, often you don't, you know, do you want your strategy betraying all markets or just certain market regimes? So that's another important part.

Uh and then you know the ranking of your signals. When you have too many signals, you know, it's important to understand how you rank them. Uh of course always important is position sizing. You know, how do you well what kind of position sizing are you gonna do? Uh very similar to that is you know how many positions. maximum positions that you're gonna have. You know, that's also important and has very interesting effects, which we'll discuss then.

Uh and then of course your entry. How do you get into the position? Once you've got a signal, how are you going to get in? And then you know what happens if it continues to scale um to sell off? Do you want to scale in or not? And then the most important part is your exit. How do you get out? So those are, you know, very similar to any other strategy. But we'll discuss more in details each of those.

And um one thing I for the longest time I looked at was smoothness of equity curve. And now lately I've actually been looking at I would say the inverse of that. That is I don't want to see smooth equity curves. Because and you maybe go, wait, why everybody wants, you know, everybody looks for smooth equipment curves, and that's the reason.

is everybody's looking for smooth equity curves and therefore any strategy that you find that has a smooth equity curve, you know everybody else is be going to. So lately I've been looking at, you know, strategies that don't have a smooth equity curve because they're likely not to be as followed as much by other people.

uh it it's not really a bounded number. So, you know, let's say the numbers range between three hundred and five hundred. Therefore I won't look at any of the strategies that have got, you know, values over four hundred or something like that. Uh just so I'm not near those smoothest looking strategies.

smoothness will continue into the future. When you've got this very You know, if you can have an equity curve that's bouncing around a lot, but yet still producing good returns, to me, actually that seems to signal a little bit better that it can deal with more different good and bad um trades. as versus a smooth equity curve, I think that's fooling you to thinking you will always just have good trades and avoid the bad trades. Uh and if you're trading mean reversion,

You will wake up one morning with a stock that's down fifty percent. Um, you know, it's you know, unless you're trading SP one hundred or five hundred stocks, but if you're trading a much larger universe, I can guarantee you that's gonna be one of the things that happens to you. Um and um Yeah, that's one of the hard parts of trading meme reversion. Mm-hmm.

And, you know, if you're trading SP five hundred stock, you're not gonna have a s that stock likely go up five, ten, fifteen percent. Um, but also at the same time, you're not gonna be the likelihood of being an SP five hundred stock that opens down fifty percent is very, very small. But if you're trading Russell three thousand stocks or even stocks aren't in any index,

Those stocks, you know, it is quite likely that it may be up 10, 15, 20%. I had a trade last week that I was up 15% in it. Um, but it's not a, you know, it's it's not an SP 500 stock. Now I can guarantee you, you know, sometime over the next year, I will wake up one morning and there will be, you know, um, it will be, I will have a stock that's down 30% or 50%. So a lot of it has to be, you know, what you know, can you deal with the that volatility, that possibility of a large straw um

a large opening gap down. If not, you know, you maybe should stay with the SP five hundred stocks. But at the same time, your returns for the SP five hundred, your strategy that you can develop from that is gonna be, you know, you may be able to get strategies in the fifteen, twenty percent annual returns.

versus if you go to Russell three thousand or stocks aren't an index, you can easily start seeing twenty-five, thirty, you know, or even higher annual returns. Uh and then if you want, you can go to some of the um foreign stock exchanges where

you even have more possibility of you know larger returns. So that is part of that there is deciding, you know, what can you handle? Because a lot of quantitative trading is about, you know, being able to push the button each each day when you see those setups.

That is one of the harder parts of quantitative trading that people don't believe, especially with mean reversion. Uh you know, you'll sa I'll sometimes look at a chart for a mean reversion trade and it's like been going down, down, down, and I'm thinking, why would I wanna buy this? And I have to not

look at the news or try to look up the news because as soon as I you start doing that, you're becoming a discretionary trader. So, you know, you just have to be able to understand that and just push those buttons and let it go.

And I don't want to see myself, you know, anywhere near, you know, a 1% of the daily dollar volume. Um, because if I'm there, you're probably going to be affecting the price. And then what's going to happen is you're going to you won't be able to do you'll have to do something like a TWAP order or VWAP orders to get in and out. Also along those lines is a minimum price.

So depending on how your commissions are set up, you may not want to trade you know stocks less than a dollar. Uh you know, I have a per share commission. So if you know if I wanted to buy five thousand dollars of a fifty cent stock, that's ten thousand shares. And using my commission structure, that's quite a bit of an expense on the commissions and that can eat up into any profit.

So that's another thing one needs to be looking out for. Um, you know, those low price stocks tend to be really good me reversion trades, but they also tend to eat up your uh commission. So yeah that's another thing one needs to take into account.

um that's been adjusted for splits, you know, way back in the eighties when it was floated, it probably says it was like twenty cents when really it was m maybe twenty dollars at the time. So how do you account for that type of price adjustment?

No, the the farther you go back in time your back testing does goes, the more important that is. So if you're only testing the last five years, you know what, you probably don't need the split adjusted price. But if you're going back twenty years, uh you probably really do uh really do need that.

at the top um that you you really only trade end of daytime frame. So is that what we'll be talking about today in that kind of context? Yep.

The meat and potatoes part is how do you measure mean reversion? And then there are, like you said, hundreds of technical indicators out there. Um and they all, you know, especially you can use a lot of them to measure mean reversion. Um My, you know, the one I popularized with Larry Connors back in the um, you know, t over a decade ago is, you know, the RSI, the two pair RSI is a a really good one for the short term indicator.

It's uh you know, it it still works quite well. I'm I I'm always surprised when I do research that how well it still works even though it's been popularized for a long time. And then that's a very common one. But there's a lot of other ones that people don't think about. Uh one is using Bollinger bands in what's called percent B. Uh percent B basically takes Where your clothes is in relation to the top bulleture bad and the bottom bullets are bad.

Uh if you're closed at the top Bollinger band, the value of percent B is one. If you're closed at the bottom Bollinger band, the value is minus one. If you're closed right the m moving average, the value is zero. So one good method um that I've used before is saying

that the percent B value is less than let's say point minus point eight for two or three days in a row. That's another good way of measuring mean reversion. Uh so you know those are two, I'd say two of my most popular ways of doing it. Okay.

How much of a pullback are you looking for? You know? And that's you know th that's the way I do it with the Bollinger bands.

Or you make them very loose and you're just getting too many trades. So as as with any indicator or any strategy, you you're always playing a balance of how you know how often are getting trades. And especially with mean reversion, mean reversion, you know, especially doing quiet pairs like th now can be can be very tempting to loosen up the parameters such that you can get enough trades. Uh that we'll talk a little bit later about market regimes.

about how the importance of that.

So I would say okay, if I'm more than four percent underneath the five day moving average, I'm in you know, I would consider I'm in a pullback. Or you can also as as with any of these indicators, you can do it as a multi day. So I'm more than four percent below The five-day moving average for two days or three days in a row, then that's your part of your setup for meeting meeting the mean reversion. Does that one uh make sense to you?

it never seems to add enough for me to justify the complexity. But again, that's something um, you know, each individual needs to decide and say, okay, I wanna account for the volatility of the stock and, you know, the percentage below is dependent on the volatility of the stock.

Another one I use is very simple, is the rate of change. Basically, you know, how much has a stock dropped over the last three days or five days? Uh again, uh I tend to do s use a fixed amount. So I'll say something like, you know, the stocks dropped. uh 15% over the last five days. Uh but again, you can make that volatility based. Uh and that's a very simple, you know, that's a very simple, obvious one. Uh it's one that I do use.

It's uh it's like I said, very simple one. A lot of these meme versions ones are also highly correlated. So, you know, uh but you know, that's that's how I use rate of change. Any questions on that one?

It's it it's um you know, it it may be a perfect mean reversion. It sometimes takes a while for the RSI or the Bollinger Band or any of the other indicators to get low enough to in to trigger a mean reversion trade. And usually it's the three to five days down is usually what it's looking at.

So the next one is uh what I uh call a a multi day closing range. So I I look at let's say the highest high over the last uh five days and the lowest low over the last five days. And I see where the close is. And normally I I want something like the close to be in the bottom ten percent of that range. And that you normally tells me, okay, you know, it's that's another way of looking at for a sell-off.

Um typically I do it in the either five days or ten days. And, you know, the values I'm looking at is usually ten per s you know, the bottom ten percent uh of that. Yeah.

Yeah. The next um another n very simple idea is just looking for uh an end bar uh closing low. So Um something like a seven-day closing low. You know, something that simple. It's like, okay, it's the lowest low or lowest and closing low the last seven days. And that could be also another great confirmation of a closing range. It's similar to the closing range, but it's, you know, this in a sense is asking it.

In some respects the clothes, you know, at zero. Uh that that's another pop uh another good one to use. So the next indicator is um the Connors RSI. I think I'd be remiss without mentioning that one. Um that one's already got set parameters. That one Um with the set parameters, it does a really good job of uh finding mean reversion traits. I'm always um quite surprised. I'm doing a little research right now on it right now, and again, I was surprised at how well it's been holding up.

Coded that up and gave it a a try. And yes, it it does seem to work pretty well also as a mean reversion indicator. So, you know, if again, you know. always be on the lookout for other indicators. Um you never know or even current indicators are popular you can turn around and use as a, you know, as a mean reversion trade. You know, moving average, it's not obvious how to use that as a

as a mean reversion indicator, but you know, once you realize, oh, you can look at a percentage below that, that becomes a mean reversion indicator. So again, I wanted to look at the L indicator and that one's definitely um has some good potential too. Yeah.

You know, something I would like to use is something like an RSI and a days down. So you know, sense of the days down is confirming, you know, it's helping to confirm the RSI. You know, you've got two indicators confirming that it is a mean reversion trade. So you know I normally will combine uh two indicators together uh to kind of give me confirmation that the that there is a mean reversion trade. But you can also kind of do

um something slightly different. And it's due and I've done this also before, is where I take two indicators and I just I ch I tune them such that the Signals from them are a very stretched trait. So I'll look for a very low RSI value. So let's say RSI value less than one. And a very low, let's say, percent B value. So a value, you know, a couple days under minus one. So both of those occurrences don't happen very often.

But when they do, they tend to be quite good trade. So I'll I'll do them as an, you know, I'll it's one or the other happening. So not necessarily they're confirming each other, but the, you know, I know when these happen, they have to be very good good ways, good trade.

So those are just two different ways of combine them. You can either do this confirmation or you can make them very strict and just say, Okay, I know these don't happen very often, but when they do happen, they tend to be good trades. So there's those are two ways to combine them. I wouldn't not combine, you know, three maybe

Four or more? No. I think you that's you're just you're getting too having too many rules at that point. You know, I I would stick at one or two, uh, you know, pick the one or two that seem to make most sense for you.

And it's a matter of, you know, having a little bit of confirmation. When you have two of the indicators, you know, giving you a buy signal, that tends to be a little bit better uh confirmation that this is a good trade. Or like I said, or you just Tune them such that they're very strict uh and give you very rarely give you trades. But when they do, they're good. So you can now do two of them that way. Um

And you have a closing range like in the bottom 10% and the bottom 25%, I found quite often really helps the returns. Another thing to do is your one-day return. There's two ways of doing this, just saying, you know, today's return is, you know, it's gone down more than two percent or three percent. But a way I like to do even better is to say is something in uh Amy Broker, which is a backtesting platform I have, have something called a percent rank.

So I take today's one day return and compare it to like the previous hundred days and I want to see it as being one of the worst, you know, of worst ten oh in the worst ten of those hundred days. Does that make sense there? Yeah. So, you know, I'm just looking at today's down day is one of the worst of the top ten of the, you know, of the last hundred days. So those two concepts really

give you that intraday, kind of like gives you an extra added boost of mean reversion.'Cause you gotta remember, part of mean reversion is you're dealing with psychology of, you know, people out there finally giving up and saying, I I want out of this stock.

It's less likely to be happening. Um also you like to use ADX. Um something like a uh five or ten paired ADX just to make sure that's good it's got good strength with it. Uh but again I try not to use too many of these other filters and again use your own favorite filters. These are kind of things that

Extra little things that I'm looking for to tell me, you know, these are the type of stocks I want to be trading. Uh other things I'll sometimes use is uh historical volatility. You know, sometimes I want a stock to be very you know have some volatility. So I'll say the historical vol the hundred day historical volatility is above forty or fifty.

Or sometimes if I'm doing a strategy that I don't want to have lots of volatility, I'll do the reverse of that. And I'll say I want the historical volatile to be less than forty. So again, these are additional little filters that I'm looking for. I try not to have too many here. Again, the more rules that you have,

the more likely that you're gonna s you know, you'll walk into the curve fitting overfitting area. Um but you know, understand why you're adding a rule and you know, that's you know, I I do use some other filters. Yeah.

to be to be larger such that it says, look, the stock is moving. So again, you know, you don't you want a stock to be have some movement. You know, you can have a pullback on something that's going very you know, but is very shallow and it's not gonna have a very big bounce.

So I personally tend to be on the stringent side and understand that, you know, at times like now, you know, I you know, I just don't see that many setups and it's, you know, that's just part of the sh the strategy. And I have other strategies that are, you know, that deal better with the low um with the lower volatility time frame. You know, and that's the important part of any strategy is understanding

the market conditions that it does well in and what it doesn't. So, you know, you know, I like I said, I prefer the stringent, but you know, some people would like the looser just because, you know, they, you know, they they they want to see a little bit more trading.

Uh position sizing, always very important how you do your position sizing. Um similar to position sizing is just how many positions you can uh maximally take. Uh that actually is something that's not discussed quite often, but there are some interesting um reasons to consider if smaller versus larger number of positions. Then we'll be talking about our entry, you know, how are we getting into the uh position. We'll talk about uh scaling in, whether that helps or not, or how

you how different ways that changes your results. And then of course, how do we get out? Our stops, exits. And of course, you know, how you know that's the end of the trade. How do you get out? Yeah.

In the meantime, if you do have some questions on mean reversion that you'd like me to ask Cesar in the second episode, I've set up a page where you can submit questions for him, however, it's only going to be open for a very limited time, so you need to get in quick before I shut it down.

So if you'd like to submit some mean reversion questions, head on over to the website better systemtrader.com, click on this episode, which is episode number one two seven, and scroll down the page you'll see that there is a box to submit questions. Now I do recommend that you do that right now. I'll be travelling through the States the next few weeks, so time is limited and I'll be shutting down that form pretty quickly. So

Get your questions in right now and we look forward to hearing them. So we hope you enjoyed this episode and look out for the next one with Cesar. We've got much more mean reversion goodness to come.