The Derivative - Beyond Traditional Trend: Leveraging Experience, Short Term, and Crypto with Mike Stendler

Episode Date: August 7, 2025

In this episode of The Derivative, Jeff Malec sits down with Mike Stendler of O'Brien Investment Group to explore their history and evolution of trend following strategies. Stendler shares insights into their innovative approach, blending traditional trend models with machine learning techniques across multiple time frames. They dive deep into the evolution of quantitative trading, discussing everything from the O'Brien family's century-long history in commodities to their latest strategies in futures, including a unique cryptocurrency trading program. Learn how modern quant managers are adapting to challenging market conditions, diversifying their approaches, and seeking alpha in an increasingly complex financial landscape. SEND IT!Chapters:00:00-00:47= Intro00:48-10:16= From Stock Markets to Futures: Mike Stendler's Investment Journey and the O'Brien Family Legacy10:17-20:19=Machine Learning Meets Trend Following: Diversifying Quantitative Trading Strategies20:20-31:19=Crypto Futures and Short-Term Machine Learning: Expanding O'Brien Investment Group's Trading Horizons31:20-39:01=Market Memories and Mentorship: Navigating Financial Volatility from the 1987 Crash to Today39:20-51:12= Gunning it back to the 1980’s, listening to the squawk box and Quantitative Trading insightsFollow along with Mike on LinkedIn and O'Brien Investment Group, also check out their website for more information at o-big.com!Don't forget to subscribe to⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠The Derivative⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, follow us on Twitter at⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠@rcmAlts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and our host Jeff at⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠@AttainCap2⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, or⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠LinkedIn⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ , and⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Facebook⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, and⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠sign-up for our blog digest⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠.Disclaimer: This podcast is provided for informational purposes only and should not be relied upon as legal, business, or tax advice. All opinions expressed by podcast participants are solely their own opinions and do not necessarily reflect the opinions of RCM Alternatives, their affiliates, or companies featured. Due to industry regulations, participants on this podcast are instructed not to make specific trade recommendations, nor reference past or potential profits. And listeners are reminded that managed futures, commodity trading, and other alternative investments are complex and carry a risk of substantial losses. As such, they are not suitable for all investors. For more information, visit⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠www.rcmalternatives.com/disclaimer⁠⁠

I'm not actually a believer in that term. Trend following is a first responder. It's kind of a second responder. And so the trend following models that we have in place are not going to kick in the day something occurs in the market. It really needs those prolonged trends. Welcome to the derivative by our SAM alternatives. Send it.

Thanks for having. us on Mike Stenler, managing director of O'Brien Investment Group. Hey, Mike. How are you? Good to see you. Jeff, good seeing you again. Thanks for hosting the podcast today. No worries. We've crossed paths at a few different events. What was it, Talking Hedge? Austin and Nashville last year we down there yeah that is correct and we'll be a talking hedge again in Nashville this fall I'll be there as well I don't think after I left you quick sort side stories but I left you at that Nashville one with my suitcase and someone's like hey let's go to one of these little honky talks and grab a quick beer on your way to the airport I'm

like fine but I have my suitcase so I get it's fine so I two bars are like no you can't come in with the suitcase. The third one's like, sure, come in. I wheel it into the bar and the music stops. The singer's at the front. She's like, look at this city slicker. I had to blazer on. She's like, look at this city slicker with the suitcase and his blazer. Music stopped. It's a little embarrassed. But anyway, we'll go to one of the honky talks down there again. There you go. Maybe you thought you were starving artist. Exactly. So give me a little personal background and then we'll get into a Brian investment group. So how'd you get started in all this crazy world?

Sure. I started my investment career, I guess, unfortunately, 40 years ago now, and started working at a brokerage firm, later went to work at a small cap value shop. And in 2003, I moved to the quant trend following side of the business. So it's been 22 years on the quantitative trading side. I never knew that. You were a stock guy? Yeah, small camp.

You were on the other side? I was on the other side, correct. In 2003, again, got into the managed future side over. I met some of the O'Brien family members throughout my career. Talking to John and Mike Durkin and some of the other family members is kind of the origins of O'Brien investment group, which, again, started in 2016. Got it. And so you've mentioned the name a couple of times. So let's talk about who the O'Brien family is and what the O'Brien Investment Group is.

Sure. The O'Brien Investment Group was started as a commodity trading advisor. It is part of the O'Brien family. That's the ownership on it. We focus on trading, quantitative trading of diversified portfolio of futures. The O'Brien family has a long history in that. The origins of that go back 100 plus years with the of R.J. O'Brien. That's been kind of the family business recently has been acquired. The family, the legacy family even goes back to John O'Brien's grandfather, Robert O'Brien, who was the two-term chairman of the CME. In fact, he was on the Tonight Show in 1964, if you kind of dig through the artwork or our PowerPoint presentation. He was on the Tonight Show with a live cattle introducing live cattle features. Then it was called the Merck, but it was launched on the Merck. So again, a long, long...

We think that's on YouTube. We could go pull that up. I'm guessing you could find it. All right. We're going to try and find that. Put it in the show notes for everybody. It's in a couple of our PowerPoint presentations as well. Who was the host back then?

Nineteen, Joy Bishop. Joey Bishop. Yes. Colbert just got canceled, right? But like those tonight shows now, they would get canceled if they brought an FCM person onto the show. Yeah, I guess if you're bringing a live cattle because you've kind of been plying that, this is live cattle. But eventually it won't be live cattle anymore. Exactly.

That's correct. We just did our, you're following the podcast last week as meat, trading meats and cattle and all that. So good timing. But before that, a hundred years ago, the O'Brien family was involved? That's correct. So I work with John O'Brien, but his father, his grandfather, his great-grandfather, we're all part of the R.G. O'Brien at futures clearing firm headquartered in Chicago. So it's, again, the long, long line of people.

And what's, as they must have been one of the founding members of CME are in there pretty early? They were, they were, they're early, right. And it was especially in what I'd call the Midwestern type commodities, and that includes milk and eggs, but also all the grains, and, you know, they've seen the evolution of all the trading of that from the old days in the pits to now, of course, it's all electronic. Yeah, I think, right? Wasn't it originally the Chicago Egg Exchange or something? I think its first iteration was something to do with eggs, which is ironic with egg prices in the news, and nobody really trades egg futures. So, and then when you flipped from the stock side to

the future side, that was with Clark Capital? No, no, that was a pre-dated that, sorry. That was another CTA that I was working with. And again, that was quantitative trading. It was Chicago. So that kind of started my career working on that side of the business. And what do you think coming from the stock side, a bunch of crazy Yahoo's or you were intrigued? It was, it was intriguing.

I'm going to talk about this maybe in a little bit, but one of the things going back to, you know, when my early, early, early days of working was the crash of 87, and again, I'll talk about in a second, but post the crash of 87, I was kind of young in the business. I started hearing about managers that had made money on that particular day. And it kind of stuck with me, kind of an intriguing part of who can make money when, in that particular case, the crisis is happening. And that was always intriguing. And as I started exploring different managers and and strategies, etc. I start saying, yeah, there's a group of managers, commodity traders out there that can trade both long and short, the stocks, the currencies, the bonds, commodities, etc.

And I found it always intriguing as compared to being on the long only side, which is kind of a one directional type trade, correct? Yeah. And did you know of the term managed futures, where you're like, I'm going to go check out managed futures, or you were thinking more commodity traders? No, it wasn't necessarily commodity trading, but it was, and I didn't know the term managed futures either. We made that up, maybe somewhere around that era. Some managers, you heard about some managers that were making money, and then particularly some of the European managers that were doing some interesting investments in, you know, kind of driven out of Europe, kind of the old

HL group that, then a bunch of people spawned out of that, etc. So it was, it was interesting. And as I kind of pursued some of that, you know, I found it said, this is, this is more interesting than it is, just being on the long side of trading. And then how, so I didn't know that. You were there at the beginning, yeah, of O'Brien Investment Group? Correct, yes. And that's different from the family office. So the family office said, let's start our own manager as well.

Well, it's, yes, completely different than the family office. The overlap is that they are the ownership of O'Brien Investment Group. And my colleague is John O'Brien, Jr., who I work with closely at the O'Brien investment group and of course he's part of the family yeah um and was that weird at the time of like hey let's start a cta when we're other ctas like hey you're competing with us at the fcm or was there any weirdness to that or you were kind of isolated and insulated enough that it it worked out no no i think the transition probably occurred prior to the start of o'brien investment group uh they had acquired clark capital um clark capital management again another chicago cta that was run by michael clark clark

long line of being a commodity trading advisor going back to the 1990s. So through their acquisition of that, which, of course, came with people and systems. And so that transition from them being a family office and O'Brien and the R.J. O'Brien being a FCM, they kind of already had that transition occur. But I think they wanted to kind of start something that was kind of new and different and use some of the legacy stuff of Clark Capital, but kind of go in a different direction moving forward so that was kind of the start going yeah i've actually traded with probably one of the first CTAs ever put client money with was clerk capital back in yeah oh two oh three ish um

and our frustrating thing with him was always he would launch a bunch of new programs that had different portfolios and like oh now this portfolio is working this portfolio is working so there was yeah seemed like good models in there but i was always frustrated with the portfolios but he got out of the business He sold that to a Brian investment group, and you guys just basically took the IP? Yes, correct. I mean, it was an ongoing business, and so we were operating the ongoing business. Michael was going to retire, and so we kind of took, you know, it took it over. It worked out for both parties.

You know, Clark was an aggressive kind of old school, you know, trying to follow the manager. And, you know, we wanted to take some of that good part of that old school trend following, but then also kind of turned it into something, but more our own, if you will. Is he still alive? Yes. Yes. Living in retirement. So, all right, now talk about so what you guys have created since then.

Sure. We have three strategies. We first started out with our legacy, our flagship strategy, which is the quantitative global macro program. and that was available to manage account, and we also did it in a fun format. That was kind of where we started on it. That later, as we started developing newer model groups, that added into some machine learning and some other types of things.

So we later have introduced a later in the sense that we just did it this year, a short-term machine learning program. The family office had an interest, and of course the evolution of cryptocurrencies, started happening. So we ended up having a third strategy, which is, which is focused on trading CME, cryptocurrency futures, Bitcoin, Ethereum, and now more recently, Solana and XRP. So, but going, going back to the, the flagship strategy, we really tried to take what was good about the old legacy trend following systems, medium-term, long-term trend following systems, all quantitative trade, along with short-term breakout. So that, you know, ultimately, we know that works. We know that it has,

that crisis alpha built into it, et cetera. The biggest difference that the evolution in starting only five years ago was the addition of machine learning models. The machine learning models tend to be shorter in time frame and they're predictive type models. So we have three different classes of machine learning models added to it. So it really kind of diversifies away from the trend-based models and ultimately gives us a different source of alpha

if trend isn't necessary working. But we still believe in the trend side. We just want to make, we wanted to introduce a model. that ultimately could potentially make money if the trend, having big trends occurring in the market. You were prescient to name a global macro instead of trend back in the day. But you knew like, hey, we want to add some more stuff to it, we want to make this more of a macro program?

We wanted to make it more of a quantitative macro program instead of having just a traditional trend following model. Because, again, we believe in that, but we don't believe that you can do some other things on it to make it a little bit less volatile or less lumpy return stream. Thank goodness, given the past 12 months in trend for the classic, right? And the more classic, the more old school you are, the more pain you've endured. And is what I've seen across.

Yeah, I would certainly argue that trend, obviously had a good 2022, but long-term trend following hasn't done a whole lot since that has been kind of two and a half years now. And it makes it difficult for investors to kind of absorb, you know, just kind of mediocre returns. They're looking for some big trends. and it's been a little bit more challenging on the trend side. And this is the second time they've been duped, I feel like, right? It was like 08, 07,08, great returns, 09, 10, 11, 12, even lackluster. Same thing here.

22, great returns, showed it stripes now 23, 24, 25, lackluster. So, yeah, we'll fix that on another podcast. But they've been duped twice, so they're like, hmm. But to me, the reason trend works. is because you get paid for enduring those flat periods, right? Like for the investors who are willing to endure those flat periods can get paid out on the outliers. Yes, yeah. The thing about trend is that, again, I threw out the term of Crisis Alpha,

which is kind of what came out of the 2008 correction. I'm not actually a believer in that term because I don't think trend following is a first responder. It's kind of a second responder. And so the trend following models that we have in place are not going to kick in the day something occurs in the market. It really needs those prolonged trends. But what I say we're believers in it is that we know that history has proven to us that there's going to be some period of time where, you know, Crudeau is going to move 40 or $50, could be up, could be down. We know that, you know, gold's going to go to $5,000 or $1,000, you know, whatever that might be.

But ultimately, there's going to be a big trend or multiple trends in the marketplace where the long-term trend models continue. take advantage of it. It's that sometimes is that you have periods of time where it trend doesn't necessarily work as well. And that's a challenging part for investors. Yeah, which is the billion, million, billion dollar question, whatever. How do you balance those two things? Everyone knows they need to add things to trend and make it more livable experience, right? So those drawdowns aren't as sharp. But how do you do that without, you know, throwing the baby out with the bathwater, so to speak? How do you lessen those down periods without taking away the big outliers to the upside? And our approach has been, yes, we have the trend, but we want to make sure we're diversified.

I think just in diversifying itself helps out. A lot of the big trend followers, because of their size, are really forced to do a lot of big financial markets and less in the commodity space, particularly smaller commodities. So, you know, we want to make sure that we have the exposures to the coffees and to the soybean meals and to the cocos and rubbers markets like that. So that if indeed there is a movement in some of those, hopefully we can capture that. So diversification certainly helps out a lot, particularly in the commodity side. I think that's where you kind of first start. But then you also want to look at multi-time windows as well. So we have medium term, long term, we have short-term

breakouts. And again, all the machine learning models tend to be on the shorter side. So hopefully that's not taken away from when the big trend occurs. And I think probably to some great, it will, but it's certainly smoothing out the return stream that you have. And I think that's the big benefit there. But ultimately, if you do have a trend going on, we still want to be able to capture that. And what, did you guys consciously make that way? Like, hey, we're adding these three timeframes, these three machine learning. It's bringing the return down, but it's bringing the drawdown down much more or something like that. Right. So the MAR ratio, if you will, is going up, even if the return's going down. Right. So, I mean, every time we have introduced a,

a new model. You know, first of all, it's back tested. Sometimes it's been traded with prop capital. But ultimately, we'll take a look at it. And we look at multiple time windows going back to the last 25 years. And we could certainly go back further than that. And then we kind of slice and dice and say, what did this model do for this window of time, this window of time, this window of time, and then collectively. So what did it do to the absolute return of the portfolio? And then ultimately, what did it do on a sharp ratio, so risk-adjusted basis. And ultimately, you know, if we're looking at various different new models, we'll say, okay, is it benefiting both of it?

Because we definitely don't want to sacrifice absolute return to have a better sharp ratio because we kind of want to have both. And so it's a small methodical process. Can't eat sharp, as they say. But, you know, once a model is introduced, it doesn't just stop there. Obviously, we're monitoring closely and we'll take a look at it. it. And, you know, sometimes it didn't be a situation when we say, hey, we're going to put down the watch list. And sometimes it could be a little tweaking. You go from a one-day bar

to a two-day bar, something along those lines without getting too wonky in this conversation. But won't walk away. We like it. Yeah. But that's kind of what we're to look at. And, you know, push come the shove, we might say, you know what? It doesn't seem to be performing as good in live trading as it did kind of back-testing training. So, right. But I think the other part, Jeff, is that we were very slow on our risk budgets. So every one of our models, we have 45 different quantitative models, 45 plus different quantitative models. Each one of these models have a special risk budget. We're divided in the 13 groups. So if we introduce a new model, we're going to start with

a low risk budget and kind of see how it works. And then if it seems to be doing exactly what we think we might bump up the risk budget on it. And again, even if we're bumping up a risk budget, it might go very small incremental. So we don't want to be introduced a new model. That's just dramatically changes the overall results of the overall program. And what, say those numbers again, 45 models? 45 plus models. Yeah, 13 different, we have, we have them divided into 13 groups. So we have multiple medium trend following.

We have multiple long-term trend following. And then in the machine learning, we have three different classes of machine learning, and then those are kind of separate model groups, and then different time windows mixed in there. So it's a lot of different models that are operating on a daily basis. So these are all quantitative. And then each of those, and I want to get into the machining learning in a second, but each of those is doing upwards of, what, 60, 70 markets?

Yeah, our universe of markets that we trade is 70 plus. And, you know, in general, you know, some of the markets are big notional exposure, big margin requirements. So you have a little bit less sizing going to it. But yeah, they. Platinum. Something like that. Yeah, I would say on any given day, we'll have 40.

45 to 50 markets on. Got it. I was getting it. You could have all 45 models across these 13 groups could be operating on any of the 70 markets. Or you're saying some of them might be too big notionally to. Yeah. Yeah. There's some nuances to each one.

But if a market is moving in our favor, it's not uncommon to have, you know, 12 or 13 different models working for us. Now, we do have a risk overlay on top of that. that will size a position. So if we're getting too many models kicking in and building our position side, then we have an overlay that will actually shrink the next trade. It might be down a half, and it could be to zero, too. Yeah.

But I like that model because it's a voting machine, basically. Hey, all of these models are agreeing on the same thing. They may all be wrong, but unlikely. And unlikely they all be wrong at the same time. So the machine learning piece, interesting for you to call it machine learning instead of AI. Did you predate the term AI? I think we weren't cool back then, so we just called machine learning.

We started our first model that we introduced to the program was about five years ago. So that was before everyone was calling it AI. I guess if we probably did it today, we'd probably call it AI to be cool. But, yeah, we, machine learning models, that was the first one, the first class of models that we introduced. And again, we've introduced a couple different classes now, three different classes in total, and again, multiple time frames on it. So what makes it unique, and I think what people, in a traditional trend following and use kind of a 200-day moving average as something simplistic, if you want to call it that. But, you know, if the market is above the 200-day moving average, you go long. If it's below, you go short, kind of a real simplistic.

you know, quantitative type trading. In machine learning, it's substantially different. You're using predictive windows. So you're kind of looking at what's the return over some period of time. So let's use 10 days as an example. So what's the return over a 10-day period of time? And then you can kind of rank it according to the returns, right? And so you kind of start looking and saying,

what were the returns for this period of time on this ranking? So you're ultimately looking, saying, what 10-day window was positive and then once you have that result, you can look backwards over the last X number of years, 25 years, 10 years, whatever you're looking at, and say, what happened during that, the time preceding that produced those results. So it makes it, it's a completely different look versus saying it's above a moving average or below moving average. I think the other thing that is unique is that trend following, a traditional trend,

model would say, I'm going to go long crude oil at 70, I don't have a stop at 62. And if the crude crude oil moves up, I'm going to just keep moving to stop up. And I'll have that trade on. It could last for a day and it could last for two years. And machine learning, we actually have to revisit that window. So if we're doing 10 days, if that 10 days is up, and let's just say that market didn't do anything. We're zero OTE. We would then have to run the model again and say, should we have that position on again? And it could get us out. We could be at a positive. It could be at a positive gain, could be at a loss, or even flat. So it's almost the antithesis of trend at extreme levels, right?

Because you're like, we've all done that being trend followers of why it feels so wrong to buy it here, right? It's come up so far, so fast. It feels wrong, but the model's telling you to buy it there when it feels like, yeah, the machine learning could be like, well, if you bought it here, nine out of ten times, that actually wasn't profitable. So your intuition was right. Yeah. The problem is the tenth time is insanely profitable.

right and it's this huge outlier is why trend does it but right put in the both together yeah even with that though all of it even the machine learning models all of those stuff stops in place so we we want to make sure that we're always protected on even on those models and then are you letting it do like how are you informing it so all how many machine learning models are there 13 13 so all you started with one and you've been adding incrementally that's And then you say, okay, this one is looking at this way. What if we tweak these three inputs basically and said look at it a little different way? Or is it all time-based?

No. So the first class that we introduced, we had a certain time frame. The model was working well for us. We introduced the same class of machine learning with a different time frame. Once we introduced another machine learning model, then we said, you know, let's take a look at a different time frame as well. Same with the third class of it. So they're completely different, their neural network, deep learning type machine learning models, but we want to have a little bit different focus to each one of them.

The ultimate goal certainly is that there's a low correlation, and the correlations between many of them are less than 50%. So ultimately, that's kind of what we're looking for as well. We don't want to keep introducing machine learning models that, you know, they could have the same timeframes and then you have a correlation of 0.8 or 0.9, that doesn't add a whole lot of value. And then those are relatively, they're short, how short term are we talking? Like you're in and out of those trades inside a day or a couple of days? No, no, they're not that short.

We're really kind of focused on kind of one to two week would be what we call short. And I know there's periods of time in the market, and this year's probably one of them, given all the announcements coming out of the Trump administration, which makes it a little bit more challenging. You're probably better off trading in one hour increments now, but ultimately we set it up to trade in kind of one to two weeks, because some of them are longer than that, but kind of one to two-week time frames.

And has it gotten easier for you guys to run and test that with the boom and AI and all these different models that you can use now? Are you just still using your same systems you use to build it in the first place? We're using the same systems on it. Yeah. It's funny to me the whole AI thing.

I'm like, people, yeah, trend followers, quants have been doing this machine learning, basically for years and years, right? I think the difference now with AI to me, little sidebar is the, they're using text, right? It's basically doing what trend followers and quants have been doing for decades with numbers, with prices, but with text. So that is different, but in terms of like putting it onto a quant model, unless you're just letting it unsupervised say, find me some any way to make money. And it says you should trade only on Tuesdays or something, right? which you're not doing anything of that sort of like unsupervised tell me a way to make money with

this data. No, 100%. We're not using any of that. Keep them on. When we're looking at it, it's strictly the price side of it. So we're not using any large language models. There's a lot of it really interesting ideas floating around on large language models. And I think as time goes on and probably faster than both of us would imagine, there'll be some interesting large language models that we could probably tap into that might provide some interest you know some some trade ideas that would be just pure price you know right now we haven't seen anything or anything that's reliable on it there's been some interesting studies on it i think it has worked out in many cases it's worked out for a day uh that's that's that's driving sentiment uh the sentiment

of the price of of gold or crude oil or whatever it might be um but it seems to be very very short-term in nature it doesn't it can't it doesn't necessarily produce the sentiment that's going to drive something for a week to three weeks four weeks five weeks so but you know i i have no doubt that we'll be looking at at some of that stuff over the next couple years my uh pet theory my worries we're going to write the big back testing problem in our world is don't curb fit don't trust the back test make sure you have all the things like over optimized don't over optimize um my worry is like all of this AI is we're basically creating a new generation of optimizers, right? That they're going to over-optimized, they don't know those lessons of don't curve-fit.

So all of this large language model is basically curve-fitting answers and text and copy and going to make movies basically that are all curve-fit, whether that has as much pain in the real world as in the financial world, who knows? Actually, the interesting note to that is that when we first started looking at the machine learning the research was curfitting, right? It was say, let's do machine learning for soybeans. Let's do machine learning for the euro, et cetera. And we started doing that.

And it worked good in back testing, but the more you kind of dug into it. So if you looked at a three year or five year window, 2002 to 2005, whatever window you picked, and it worked really well. But you could curfitting it. But then when you started looking at different timeframes, the performance wasn't there. And ultimately, when we started introducing, and we said, you know what, if they're more successful when we weren't curfitting, and we laid one single model

across all of the different markets that we trade. So you got away from the whole curve fitting. So that what you just said is 100% positive and still works today. Get away from the curve fitting and some models on a specific market. So, you know, sometimes it works. Hey, it works really good for two weeks and then doesn't work for you anymore. My words, like the people are going to get so used to using the AI. They won't know when it's curve fitting or not.

Right. That's another podcast topic. So this worked out so well that you guys separated out as a separate program or it was doing some. It was different enough that you said, hey, let's separate it out as a separate program. Well, one, it was working well for us. So that's first and foremost. But secondly, because we took us the shortest of the time windows and we thought we were diversified enough, but we wanted to focus on kind of the short-term nature of it versus

the longer-term side. So we took and introduced a new program. It's called short-term machine learning, and that's all kind of one to two weeks. We don't go anything longer than a two-week time window. One, it's all machine learning, but two, it's a shorter time frame. And ultimately, we were hoping that would, obviously, these models are all used in our quantitative program, so it's not going to be completely different. But ultimately, we wanted to have a little bit of more diversification and hopefully lower correlation between the two strategies. And, of course, it's less models that are in it as well. We could offer the program at a lower minimum. And again, I think the machine learning makes it kind of special and unique as well. They're both diversified.

So that's definitely a positive. Is it, so what, it runs super low correlation with trend, not just with your trend, but overall the trend indices and whatnot? It runs a low correlation to trend, yeah, but it's going to have a little bit higher correlation to our flagship quant program because all the models are being used in it correct. But yes. And I would expect that if we have a really big trend, some of the models in the short-term program will pick some of that up, but we're not going to have a, if a 2022 happens again, these models will probably, they're certainly not going to produce that dynamic returns that a trend manager will have. Yeah. But that's what you want out of that space, right? You're like, hey, grab as much of that

upside as possible and as little to zero of the downside as possible. Yeah, because all these models have relatively tight stops besides, what I mentioned before, about revisiting the model and running the model, so they might be getting you out. But ultimately, the stops are modestly tight, certainly compared to a trend manager. So, yeah, if you have a big trend reversal, you'll be out relatively quickly. last piece you got into crypto so that's a separate program and how did that come to be have you always been a crypto guy on a personal basis no yeah but maybe that's a demographic it could be a level of interest but so the family office started investing in some different crypto projects not necessarily specific spot

cryptos, but in various aspects of the crypto industry. And we started having conversations at our firm really started probably back in 2018. And we started looking at the crypto market. Obviously, it was evolving, and we'd have these huge runs, and then you have these crypto winters and that. And so we said, you know what, let's, there'd be some interest there if we started looking at the crypto space. The evolution kind of helped the decision process as well, Because along the line, it wasn't just spot cryptocurrencies. The CME started trading Bitcoin futures, later Ethereum, and again, most recently, Salarium and XRP.

Our level of comfort with counterparty risks with the CME skyrocketed, we were futures traders. That's the origins. That's our history of us, right? So we felt very comfortable trading the future side of it. When we first started, we started the fund in January 2021, and ultimately we were doing both the futures contracts and some of the spot a year later we just said let's just trade the futures on it um you know it's keep in mind this is this is cryptocurrency so it just by nature it's a lot of volatility to it yeah then if your diversification is bitcoin ethereum and you know a couple

other cryptocurrencies you're kind of lacking diversification they kind of trade together right so you have a lot of volatility to it it's a it's a fun and interesting fund it's done well um but it you know to some degree and it's long short right so we can traded, but it takes a special investor to want to look at a cryptocurrency, but there's a lot of dynamic returns to it. One thing about cryptocurrency, if you're, if a quantitative trading, one of the things you want to like the boast is something that moves a lot over a window, over periods of time, has a lot of volatility because that's where the opportunity set is. And then real quick, back to the trend and the machine learning. Do they include the crypto futures

in their universe or it's only in the crypto program? We actually do trade Bitcoin, the microcontract in our main program. It's one of 70 markets, though. Yeah, yeah, yeah. If a individual investor that wanted to manage account with us, I said, please don't trade it. We would be happy to take it out. It's not a big deal.

It's one of 70 markets. So the impact on that particular program's results is going to be pretty modest. Yeah, but to me, it's always when trend followers, like, no, I don't want that in the portfolio. I'm like it's it's an asset that moves up and down like check your feelings on it if it's real or fake but I mean I guess some would say the worries it goes to zero one day just automatically but yeah we were trading it long and short so yeah exactly the opportunity to make money if I look at it this year we're down a little bit in Bitcoin trading a bit

if you follow Bitcoin it kind of peaked at the end of the year we had a kind of bottomed out in February where it's moved back up but it's kind of still you know it hasn't had a a huge move this year. Some other ones had big, but Bitcoin hasn't had this huge move. So it's been a little bit tougher market to trade. Again, if I looked at it today, we're probably close to breaking even on Bitcoin this year. But it's, again, one market. So even if it has Bitcoin goes at $200,000, we'll make some money on it, but it's not going to be the key driver in the portfolio. Right. In the main program, in the crypto program, that would be lights out. Yeah. Well, that's, yeah. So in the crypto program, is there a,

Are you trying to have a beta to Bitcoin or to any of them, or is it purely absolute return, no correlation? Well, we trade all our quantitative models, and we actually have a discretionary, kind of semi-static long position that we use kind of for risk management. So we have a little bit of beta there, but we could, in that pro, in that, it's called Pekis, but Pekos could be, we could be net short and we could be 100% plus net long. So we do have both sides of that to trade on it. But there's definitely a bias. I think if you trade Bitcoin and your investor in Bitcoin, you probably have a bias to the upside. But given the fact that we are traders at a part, we're certainly going to take advantage of the trading mentality. And we could certainly be short Bitcoin as well at Ethereum and the other ones.

Yeah, it's surprising me, I guess there's a couple out there. But it's surprising me there haven't been more pure crypto programs like that using the futures. until very recently it wasn't quite liquid enough you've got a huge role cost how do you guys handle that or because you're trading it's not really you're not trying to replicate the futures we're definitely not trading but i think your first point is probably more valid when bitcoin you know when bitcoin first started trading on the cm again it added a lot of value from a counterparty risk but margin requirements were high it was difficult a lot of the fcms didn't want to you know trade it if only you knew someone at the fcm yeah

Probably helped out. I mean, I think the FCM's dragged their feet, but eventually did it. There's not a ton of volume. We, you know, if you look at the options, it's still pretty thin. Ether's probably, you know, came along next. There's still not a ton of volume there. It's getting better. And if you look at the recent ones they've introduced Solana and XRP,

Solana, the volume's pretty small there. So, you know, if indeed you're talking about, you know, some, you know, gigantic managers stepping in and wanting to do some strategy as a stand-alone, it would be a little bit challenging. There's plenty of liquidity for us, but, you know, if you were going to talk about, you know, billion or $2 billion or $3 billion, you're going to start running in some issues there. Yeah, the allowing the ETFs has helped out the space, right? A lot of them are using the futures, and that's provided the liquidity for us for the traders. So what's next? You got more, more programs on the list, or you guys are set for a little bit?

I think we're definitely set for a little bit. We just introduced the short-term program, even though we've been trading it for a couple years, but that's new. We kind of want to keep working that one. You know, we're always going to be looking at new model opportunities, etc. We get together once a week as a group, and we chat about what's happening with our models, we, what's happening with our, with our different markets that we trade, et cetera. We're all pretty open about saying, hey, we should look at this particular market or that.

We don't trade lumber. We don't trade, you know, several other markets. So should we introduce a lumber to it? You will talk about it. Is there liquidity there? Does it make any sense, et cetera? You know, a lot of those type of ongoing conversations, most of the time, you know, it doesn't take a step forward. But I think the three programs that we have now are good.

we're not looking to expand that, certainly for the balance of the year. Yeah, and that's, we've had some on this podcast of the new frontier for trend following seems to be in these smaller esoteric markets. Maybe they're traded over the counter and those kind of things. But so nothing like that at this point. A little segment we do with all our guests here. travel back in time to any market event, just to witness it, to trade it, to do whatever with it? What would it be? And why? It's an interesting question. And I think I'm going to answer it if I knew then what I know now, it would probably be a lot more interesting. But I mentioned

I had started my career in the mid-80s. I was working at a brokerage firm when the crash of 87 came along. I was responsible for the mutual fund and annuity department. Anybody that was familiar with the business back then knew that most annuities were fixed annuities, fixed rate, interest rate annuities. And most of the mutual fund business was all government bond funds, right? That was kind of the big thing, certainly in the back in the 1980s. So I'm in my office. I'm in my office both the Friday before and then the Monday crash. and we had a little box in every office, which was in Chicago?

No, no, it was in Milwaukee. It was a little box called the Squawk Box. And if anybody was not familiar with the show in the morning at CBC called Squawk Box, we all had this little speaker box in our office because, of course, we didn't have emails and Internet and all the other stuff. And so every day, all the department heads would have to get on and talk about what's happening for the day, right? So that in place all day, you can turn an hour.

off. It kind of plays all day and people are talking about what's happening in the markets, etc. We also had Kotron machines, which was kind of before Bloomberg machines. The interesting thing about it was the Kotron machine for the Dow Jones Industrial average, I only had two digits. When the market was down more than 100 points, you didn't even know, unless you had to do the math to say, okay, the Dow Jones is Dowdx, right? And so my entire day was spent on the phone, right? There was no... So it showed it was down 99 points, basically. And yeah, no, so you look around and say, okay, it's the market's down 25 points, whatever, not a big deal, right? And because of the, what I mentioned about the departments, I was responsible for,

my phone really wasn't ringing that day. I'm talking about, you know, keeping on, Friday was down big, and then Monday, of course, was the crash. But I was only a couple years out of school, you know, kind of kid not really fully understand what's what's happening. And as time kind of went on on Monday. I was like, it's kind of a boring day, kind of slow. And I just heard a lot of yelling and screaming. And, you know, this is 1980s and a lot of explicit words, but going on, but I went over the OTC trading desk and it was just a zoo. My overall point on my story is that, you know, I was kind of just so naive about the business and what potentially could happen in a day like that. And then, you know, it was just, it was just really fascinating to

much, but I kind of really didn't appreciate it at the time of going through it of what was actually happening that particular day, you know, melting down to the market. It was, it was, it was interesting day, but I kind of wish I knew, you know, had the hindsight to say, boy, look what's going on and really kind of absorb a lot more of it. Yeah. Did you have mentors and went out? Like, hey, take every dollar you earn from here on out and just plow it into the market, or were people spooked? Oh, I mean, definitely spooked. Definitely. yeah yeah yeah it was it was kind of an interesting time because people were spooked and you'd be like the stock market are you crazy didn't you see last week if you think of 2008

I mean there was plenty of times to be spooked and yet it took another you know several months before you know even though all the kind of market was collapsing in in August of September of 2008 you know we kind of bottomed out in March of 2009 right so yeah if you jumped in you kind of you had to have staying power to be able to do it turns out you know those are are great times to buy, but, you know, unlike the crash of 08, you know, it took a while. 08, 2008 was kind of a rough year. Of course, you know, 1988 came along and things kind of went back to, you know, the good times again. But, you know, I guess if people look at the results of the stock market, SB 500 Dow Jones back then, you know, it was a good year in

2000, excuse me, 1987 from January to August. It was a really good. year. So when it started correcting, it's kind of taken profits, but it wasn't really an ugly year. And it didn't turn out to be an ugly year overall, just that it had all, unfortunately, the drawdown all happened in the course of a two-day period of time. I think that helped the futures markets, right? The whole portfolio insurance and everyone saw the flaws in that and said, hey, maybe we should edge with actual futures and futures options and options, probably helped Chicago, that event. It probably did. But, I mean, if I think back of the term portfolio insurance, a lot of people were going,

I heard the word, but I'm not exactly sure what it does. Yeah. And when the market keeps going up, as I said through the bulk of 1987, it was kind of like, I don't know what insurance you need. Everything's great. Yeah. And I actually don't even know. I got to look at it, what they were actually doing with that portfolio insurance piece.

Awesome. What else we got to talk about? Anything? No, I think we covered it in. I think as we look at this year and kind of going forward, I mentioned, I kind of touched on it before. This has been a challenging year for, I think, most managed futures, managers, regardless of if you're a machine learning trend, medium term, short-term trend, relative value, it's been a very, very difficult year. Part of it has been headline-driven. certainly the tariff announcements back in April, the budget bill, there's been plenty of kind of

headline things. I think we're slowly getting beyond the tariff scare. I mean, we're signing agreements with India and I think this morning, et cetera. So we're getting past a lot of that, and I think ultimately that's going to start putting some movement and trends and a little bit kind of less chaos into the market driven by the headlines. that's usually good for trend following. It might take the rest of the year on it, but ultimately we know that eventually the markets are going to trend.

It's not going to be, you know, we could go in recession with the all the tariffs and ultimately, or we could start seeing the dollar that just started moving down. If you look at the Euro, Euro one from 104, 105 to 118, now it's back to 114 handle. Yeah, who knows if the dollar starts weakening against some of these currencies

where you start seeing a Euro going from 114 handle, a 140 handle. Yeah. So those are the type of moves that ultimately we know that as managers we can make money on. You know, same with gold. Gold's obviously moved up a lot. It's been kind of sputtering around this 32, 3,300 level. Ultimately, we think it could make some money, but, you know, gold could certainly go to

$5,000 and that's where we can make some money on it. So, you know, those are the opportunities. I think we're always optimistic that you have to kind of have to be. But, you know, we think we can generate some big gains again. And you never know. There's always that one-off market. Again, last year was coffee, a little bit cocoa. Some of the soft markets really helped out.

We made a fair amount of money on the grains last year, being weak. This has been a bummer crop for the grains, so kind of where that goes as we get into fall. So, yeah, there's a lot of opportunity. Yeah, but I was holding guys up, right? Trend following's dead, all this bad period. I'm like, well, there's a few guys that have bucked the trend, pun intended. So, yeah, keep doing what you're doing. Those models seem to be covering the trend downside for now.

keep doing your student. And I prefer if you're a European and you say Kokoa instead of Koko. Kikoa, okay. Kikoa. I've always loved when I'm at a panel and these guys are like the Kokoa trip. And you know a lot for a quant guy, right? Most of these guys can't tell you what any of their markets prices are at. So do you actually, how do you reconcile? You follow these.

You like to take in the news, follow what the markets are actually doing, even though your model doesn't really care why they're going up or down. Yeah. It's both professional at a hobby. You kind of find out what's happening. I get up early and kind of, you know, turn on the CNBC or Bloomberg, and they're talking about the various markets. And of course, our industry puts out a lot of information. The FCM has put a lot information about what's happening on the fundamental basis. And so you can kind of hear both the fundamental side and the quantitative side. And ultimately, normally when there's a big movement in a market. It's usually a combination of quantitative and fundamental. Everything's kind of

in the same direction. So if, you know, corn is going down and the quants are shorting it and it's going to be a bumper crop and it looks like it's going to be, you know, supply is going to be fantastic. Ultimately, usually those are going to be the better trends. You know, the models will dictate. It doesn't really matter what I think if the models are dictating us to be short. We're going to be short that market and vice versa. You know, I'll have my opinion. but we don't we don't overlay our opinions on any of our trading i tell quants i'm like you need to learn how to talk the market because you're going to sit in an investor meeting they can understand that it's a quant model but they're really going to understand when you talk about there's a huge

crop in corn and that's why it's going down right like so to me my advice is always like hey learn how to talk the market whether you care what it's doing or not you need to learn it do you think you could be a discretionary trader um Yeah, probably. Probably. But, you know, I mean, you do what you want to do. I think the best, the best times is when it, again, it's a combination of quantitative trading and fundamental trading where you can kind of look at both sides of insane. Because, you know, the worst thing about any discretionary trader is I'm right, I'm right, I'm right, I'm right. And the market's going against them and they keep thinking, I'm right, I'm right, and right. So you kind of have, you want to have the technicals behind you to say, you know what? You're not right. So, but you were just talking about the fundamental side, there was. There's a FCM that put out an interview with a Mississippi cattle farmer. And he's a farmer and he's kind of saying, well, there's less cattle and supply it's down, but demand is still there. You know, forget all the everything that's just kind of somebody is saying, you know, prices are up.

And, you know, meats is one of the area, both live cattle and lean hogs that we've made some money on this year. It's been a good trade. It's, you know, two small markets. So we made some money on it. All-time home. Awesome. I think we'll leave it there. Thanks, Mike.

Jeff, appreciate it. Thanks for your time. Great talking to you. You've been listening to the derivative. Links from this episode will be in the episode description of this channel. Follow us on Twitter at RCMaltz and visit our website to read our blog or subscribe to our newsletter at RCMaltz.com. If you liked our show, introduce a friend and show them how to subscribe. And be sure to leave comments.

we'd love to hear from you. This podcast is provided for informational purposes only and should not be relied upon as legal, business, investment, or tax advice. All opinions expressed by podcast participants are solely their own opinions and do not necessarily reflect the opinions of RCM alternatives, their affiliates, or companies featured. Due to industry regulations, participants on this podcast are instructed not to make specific trade recommendations nor reference paths or potential profits, and listeners are reminded that managed futures, commodity training, and other alternative investments are complex and carry

a risk of substantial losses. As such, they are not suitable for all investors.