John Krautsack – From Turtles to Artificial Intelligence in Commodity Trend
In this episode, I talk with John Krautsack. John is the Chairman and CEO of EMC Capital Advisors.
John Krautsack directs all investment activity at EMC. He started his career in the futures industry in 1985 as an assistant to a prominent S&P 500 trader at the Chicago Mercantile Exchange. From 1989 to 1995, he managed trading operations for De Angelis Trading/Crown Capital Management, JPD Enterprises and ALH Capital. He joined EMC in 1995, overseeing trading and managing the portfolio until he assumed the role of Chairman in 2013.
EMC Capital Advisors is a leading investment management firm that has successfully managed assets for institutional and private investors from around the world since 1988.
- Principals have 80+ years experience managing client assets through all types of market cycles and periods of acute economic and geopolitical distress.
- Proven expertise in managing a diversified portfolio of global commodity and financial instruments.
- Disciplined and systematic investment methodology.
- Quantitative research process that allows trading and risk management systems to evolve and adapt to the current market environment.
John and I talk about the history of EMC going all the way back to the unbelievable story of the Turtle Traders. We touch on how EMC has adapted their trend following models over the decades. We discuss how advancements in machine learning and artificial intelligence have helped their in-house team improve their risk assessments and monetization.
I hope you enjoy this conversation with John as much as I did…
Listening options:
- Listen on Stitcher
- Listen on iTunes
- Listen on Spotify
- Listen on TuneIn
- Android users can grab our RSS feed here.
Have comments about the show, or ideas for things you’d like Taylor and Jason to discuss in future episodes? We’d love to hear from you at info@mutinyfund.com.
Transcript for Episode 28:
Taylor Pearson:
Hello and welcome. This is the Mutiny Investing podcast. This podcast features long-form conversations on topics relating to investing, markets, risk, volatility in complex systems.
Disclaimer:
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 Mutiny Fund, their affiliates or companies featured. Due to industry regulations, participants of this podcast are instructed to not make specific trade recommendations nor reference past or potential profits. Listeners are reminded that managed futures, commodity trading, Forex trading, and other alternative investments are complex and carry a risk of substantial losses. As such, they’re not suitable for all investors and you should not rely on any of the information as a substitute for the exercise of your own skill and judgment in making a decision on the appropriateness of such investments. Visit mutinyfund.com/disclaimer, for more information.
Jason Buck:
This is Jason Buck from Mutiny Funds and it’s my pleasure to sit down today with John Krautsack of EMC Capital Advisors. We’re going to talk about all things trend following machine learning, et cetera, so I’m very excited for this conversation. John, you have a tremendous storied history within the CTA and trend following space. But you started at the CME with an S&P futures trader in 1985. What got you into this industry? Was it serendipity? Did you fall into it? Tell me how you started off in ’85 on the CME.
John Krautsack:
Sure, I did fall into it. I had a good friend, whose brother-in-law was a big local trader in the S&P 500 Pit. And he asked me one summer if I’d like to go down to the Mercantile Exchange and get on the trading floor, and see what goes on down there. Because I had at that point, I had no idea what happens down on those exchanges. So, we got down onto the trading floor and it’s chaos, obviously, new to me. And this guy comes rambling out of the pit with his tie over his shoulder and just totally disheveled. And we follow him around. He doesn’t really talk to us that much. He’s counting his trading cards. We go to the merch club and had a lunch. He’s still not really paying attention to us at all. He’s looking at the ticker tape with his cards.
John Krautsack:
And then finally, we ordered lunch and we talked a little bit. And then during lunch, he took all his trading cards out of his pocket, and he threw him across the table to me, and he said, “Count my position.” He goes, “The buy side on one side and then the sell side on the other side.” And I’m like, “You’re short like 45 S&P contracts.” And he goes, “You want to work for me.” That was, I wasn’t even looking for a job. It was just, so he hired me on the spot. I worked for him for several years. A huge local. Very successful. He was on the front cover of The Wall Street Journal. He eventually wanted to take it off the floor and run a hedge fund.
John Krautsack:
So, he moved me out to Arizona to start a hedge fund with him. I managed all the trading for the hedge fund. Over time, he was a great discretionary trader, but over time, he knew the best way of successfully trading these markets is more systematically, so we really tried to implement that. We raised some
money from some of the big allocators early on in the business. And so, I lived out there for several years in Arizona. He really had a hard time staying disciplined to the systematic strategy. He eventually, it kind of drove him a little crazy, because he’d override some of the signals and do some trading overnight. So, eventually, he threw in the towel. He actually called the investors and had them take their money back.
John Krautsack:
So, it was during that period of time that I was following in the Wall Street Journal the Richard Dennis Turtle Traders. They were putting up triple digit returns, they were systematic. It just really intrigued me. So, as we were ending that hedge fund in Arizona, I reached out to Elizabeth Cheval and inquired about working with her firm. And she basically flew me out from Arizona, and interviewed me. And at that time they weren’t looking for to add another employee. So, between interviewing her, I worked for two other hedge fund strategies.
John Krautsack:
And then while I was working for a hedge fund strategy in California, she called me up and wanted to hire me, not knowing that I wasn’t in Illinois anymore. I was in California. So, I took that position and basically turned me into a… we are trading 24 hours around the clock, so they wanted someone who was cross-trained in every aspect of the business. And so, that’s how I started out at EMC in 1995. So, I’ve been at the firm for 26 years. Over time, I became Liz’s right hand man in business. All the allocators knew I was somewhat running a big portion of the business and that was reflected.
John Krautsack:
And she passed way suddenly on us from a brain aneurysm and I was the success plan of the firm. So upon her passing, 100% of the company was transferred to me. That was in 2013, so but starting the firm EMC, Liz was one of the Rich Dennis Turtles, so that was a fascinating story and experiment that Rich Dennis did with his partner, Bill Eckhardt. It was, she had a little bit of luck getting into the business as well.
Jason Buck:
Well, that’s, I wanted to kind of dive in that story a little bit, but like you said, there’s a lot of luck in life, and then a lot of skills. It’s sort of we’d rather be lucky than skillful sometimes. And I think that that’s the unique history of EMC, founded by a female trader with Liz Cheval, but also, there’s a handful of female traders during the turtle experiment. But can you kind of for maybe some people that are new to trend following your CTAs or even the turtle trading experiment, can you kind of give a synopsis of what Richard Dennis kind of the experiment they ran with those turtle traders?
John Krautsack:
Sure. Back in the early, early ’80s, Rich Dennis was an incredibly successful trader, and his partner, Bill Eckhardt, was also. They had basically not a disagreement, but almost like a trading places bet, that Rich believed that you didn’t have to have an eight skills to be a good trader. That you could teach anyone to basically follow rule-based trading, how to get into a position, how to get out of a position, and how to manage that risk. And so, they had a bet with each other and they took an ad out in the Wall Street Journal, 2000 people responded to this ad. And Elizabeth Cheval was one of the first women as part of the original 12 turtles.
John Krautsack:
So, Rich and Bill trained these traders for several months. And then they opened up accounts, whether he funded them a million dollars or a couple of million dollars, and had them based on all this training, go off and start trading on their own. Well, Rich was definitely proven right because these traders, who had nothing really to do a trading were very successful trading his strategy. And eventually, more than a handful, ended up breaking away from that, that whole tutelage and forming their own management companies, and have been successful since. So, Liz formed EMC in 1985 and the classic program was the strategy, the trend following strategy that she first implemented at the firm. And since then, we’ve had several other strategies that we’ve built.
Jason Buck:
There’s so many fascinating things in that story, in general, with the Turtle Traders. One I always think about, when I’ve responded to an ad in the Wall Street Journal for trading, it’s like it seems too good to be true. So, I don’t even know if I would have participated in that ad. It seemed too good, but maybe it’s different at that time. The second is what was fascinating to me is of the two turtle cohorts is how you had even dispersion in returns. Even though they’re supposed to do exactly what Rich said, some people maybe took more leverage, different leverage, maybe didn’t take some trades. And so, you even had differences in returns.
Jason Buck:
And then three, after they left, a lot of them added to or changed up some of the methodologies, there look backs, et cetera, breakouts, and they kind of expanded on Richard’s ideas or they didn’t. And some kind of even stuck traditionally to the discipline of the program. So, that’s been very fascinating. And then if anybody has a chance, if you haven’t heard of the Market Wizards books really popularized a lot of these individual traders that went on, whether it’s Salem Abraham, Jerry Parker, et cetera, or even Liz. And they’re fascinating to read all of their stories, especially anybody that’s coming up trading now.
Jason Buck:
And then, so a lot of times we have, a lot of investors are unfamiliar with the managed futures space. Part of that, I think, is a marketing problem on naming, right? It’s called managed futures, CTA, commodity trend, trend following. So, let’s just start off with some basic definitions. In your mind, what is a definition of trend?
John Krautsack:
Yeah, so we look at trend in two different ways. We look at trend from a momentum standpoint and then we also look at trend from a range dependent standpoint. And bottom line, both of them can be interchangeable. We’re just trying to capture outlier price movements, whether that be to the upside or to the downside. And when you have methodologies that are trend following or momentum based, and that’s what you’re trying to do, you’re going to get a high correlation between managers, because the outlier moves really, really have an influence on the correlations. We have that within our own strategies, our systems. We try to even though we’re trend following or momentum strategies, trying to catch outlier moves, we’re constantly trying to diversify those systems within the strategy. And I’ll probably go into a little bit of how we go about that.
Jason Buck:
Yeah, part of that, though, is a lot of times trend and momentum can be synonymous, but it depends on who you’re talking to. How would you differentiate between trend and momentum in your mind?
John Krautsack:
So, the way we differentiate is with our systems, our trend following, core trend following systems are more like range dependent. We don’t trade the breakout. All of our strategies have evolved. That’s where our success has come from at EMC and our longevity, is because we have a systematic way in our research process of evolving these strategies. So, for the trends following, we’re really looking at confirmation of a trend over three discrete periods of time. And so, we look at a longer period, maybe 200 days, maybe 300 days then we have a smaller period that would confirm the trend at. Medium term at 30 to 60 days, and then a shorter term parameter. So, we need confirmation on all three fronts before we can look at a possible trend.
John Krautsack:
We also have volatility filter, so we’re looking at current volatility versus past volatility that can look back as far as 200 to 500 days. That’s the optimization process. We come up with our parameters and our lookbacks through that optimization process that I’ll talk about later. Momentum are more statistically driven strategies. So they’re time weighted to more of the present, so 10 to 30 days. And we’re looking at closes to closes and statistical probabilities of a trade emerging.
Jason Buck:
Perfect. And then the other benefit to being a CTA or trading futures is the global diversification. So, I guess tell me a little bit about the different asset classes that you would trade at the EMC.
John Krautsack:
Yes, we have a tremendous amount of diversification in our portfolios, and all of the programs that we offer. So, we’re trading global stock indices, global fixed income, both short-term duration and long-term duration. We also have an expansive portfolio of commodities, so we’re trading energies. We’re trading sauce, based metals, precious metals, trade meats. We also trade currencies.
John Krautsack:
So, we have a lot of exposure, because our AUM is it’s just pressing on a half a billion from AUM, but we still have the ability to access some of these markets with a good impact. We have to weight every market, risk weight, every market, and even some of the markets like lumber. I know that a group out there wrote an article about us because not a lot of managers caught the lumber trade. And we actually caught it with a decent impact on our portfolio, even though it’s a lightly weighted market in our portfolio.
Jason Buck:
Yeah. You just touched on one of my favorite things, I try to maybe highlight in three ways is that the lumber trade, for example, is highly capacity constrained, right? So, if you’re a huge fund, you’re not probably going to have lumber on the books, because it’s not going to move the needle as far as your P&L, so a lot of people weren’t in that trade, which is interesting. The second is, like you said, at a half a billion, there’s been a trend within the CTA space for people to be large asset gatherers, right? And as soon as they start getting $5, $10 billion assets under management, they’re going to be trading a lot less
markets, because their market impact is going to be greater, so you’re not getting a lot of that diversification.
Jason Buck:
So, the third way we look at is we really like to find managers that are still trading at least 40% commodities, as you guys are, because that’s where you’re getting that real diversification that may be missing in a lot of the larger firms these days. And part of that is, somebody go, “Why don’t we trading 100% commodities?” Well, it’s not quite that easy there because if you go 100% commodities, you’re going to miss out on a lot of those short trades and a lot of the financial instruments.
Jason Buck:
Like for example, 2008, if you’re not trading bonds, or even stock indices, you can be missing out on those longer term trends on the downside as well. So, it’s a nice balance of having a good amount of commodities, a good amount of different asset classes, and still having the financial assets, but not raising so much AUM, that you’d have too much impact on these capacity constrained commodity markets. I just had a mouthful. So, if you want to push back on any of that, feel free.
John Krautsack:
Well, I like what you said, because I’ll tell you, one of the most difficult things and what I think we do really well at EMC is that we have a disciplined approach to our strategy. And what I mean by that is, we do have 40% commodity exposure and there’s been some periods of time where the commodity markets in general, were horrible. Several years of horrible performance in the commodity sector. And the fact that we keep discipline to our strategy, and don’t kick those markets out of our portfolio, and continue to tray that diversification, we get rewarded. Like the past two years, I mean, this whole inflationary environment that we have, we’re really taking advantage of keeping those markets in our portfolio.
John Krautsack:
So the diversification is, we believe that on all levels. We have diversification with our systems. We have diversification with our markets. We just believe in it. And people might think systematic trading is easy, but it’s a discipline that could make it easy. It’s not easy for most people to sit through these strategies.
Jason Buck:
Yeah, I think there’s a couple of things you said. One is even if you’re able to trade commodities, et cetera, you may be tracking 60 to 80 markets, but in any given timeframe, you might only be trends in 5 to 10 markets. This is why you need to be able to trade a lot of markets and you might be long and short, depending on the different markets. But this is why it helps if you can trade lumber, because that’s where you could see the actual breakout trend, where you may not see it in a lot of other asset classes. So this is, it’s interesting that if people don’t know about trend following, yes, you may be trading 80 markets, but at any given time, you only may be trading 10 to 20 of those markets.
Jason Buck:
The other thing you touched on is inflation. And I think the most interesting piece about that is, inflation is a very pernicious thing for all of us to cover. When we look at all the asset classes, how do we run our investment book or our savings and manage in inflationary environments. And I would argue that
possibly that CTA trend is your best guess. It’s our best bet. Meaning like there’s a there’s a high correlation amongst commodity markets to inflationary markets, at least historically. But then by riding the trends of those markets and getting in on those trades, it’s your best example of getting a high beta exposure to inflation. So, hopefully the rest of your portfolio can maintain its purchasing power parity. Do you look at inflation any differently or how do you think of the way CTA trend overlaps with inflation?
John Krautsack:
Yeah, I mean, I look at it as a whole because it’s not just the commodity markets that take advantage of this type of inflationary environment. It’s also the currency markets along with the fixed income. Our current environment here is we’re net short, short-term duration, fixed income has had a very good impact on our portfolio. So, we’re taking advantage of this environment through a big chunk of our portfolio because we do have commodity currencies like the Mexican peso and the Canadian dollar that really performed during this period of time.
Jason Buck:
Yeah, like people go, “You should just have gold for inflation,” but that’s just one path dependency. You never know it’s going to take, so you want to have as many bets as you can for those path dependencies. You also, you just brought up another interesting question I always liked to ask CTAs, so I’m kind of jumping forward here is, a lot of people would argue that you’ve just ridden the bond bull market. So, if we have rates rising in a rates rising environment, that’s I think CTAs will have a hard time at least on their collateral positions, and T-bills or et cetera. How do you view that that argument about you’ve just been running a bond bull market?
John Krautsack:
Well, it’s valid from a research standpoint. You can also say that with the stock market as well, your research and your outcomes are definitely biased from on the long side. But from our standpoint, we’re building technical systematic models, that are truly just looking at price activity. So, we’re able to get short as easily as we are to get long. So, we try to remove those biases from our research process and really just build robust systems. I could talk a little bit about that, because I think it’s important. So, what we do at EMC is we have four core systems in the classic program.
John Krautsack:
And those four systems are optimized to not just the bonds and the stocks, they’re optimized to every market in our portfolio. So once we run those optimizations and come up with the core parameters of that core logic of our system, we trade that system across all the markets, so corn, gold, S&Ps, they have the same exact system parameters that are traded across all of them. We feel that we have robust systems. If we can successfully trade those systems instead of customizing the system to the S&Ps, or customizing it to the Japanese yen, we believe that our research process is more robust by doing it this way.
Jason Buck:
Yeah. And I want to, I’m actually going to get into like those four core systems, especially with classic. But we also need to get to the research and infrastructure that underlies those kinds of systems. But I also had one more question before we get to that is, I think about almost trenching or putting together sleeves of CTA managers with short-term lookbacks, medium term, long term. And, historically, people
would say you get large dispersion between the different loopback systems that people run, and you have different market regimes are good for different systems, like in 2008, it might have been better for long term systems. In March of 2020, might have been better for short term systems.
Jason Buck:
So, without like, obviously, for compliance reason, we can’t dive into numbers or anything, but I always think about you guys more as a medium-term system. And you might argue it bleeds into short term and long term, but that allows you sometimes to do just as well in 2008 and as well as to 2020 without being on the far extremes of that spectrum. Is that a fair assessment?
John Krautsack:
Yes, it is. It definitely is. Our average holding period is probably in the wheelhouse of 30 to 40 days, but we do have systems that go, long-term systems that have an average holding period of 120 days and we have short term systems that average about 15 days. So, we really don’t want to be basket into there is a big run with short-term traders popularity and there’s big round at long term. We want to stay diversified with our markets with our systems, because we feel that’s the most successful way of not missing these trends.
Jason Buck:
Got it. And so, getting to the meat of the matter, the core thing is what you guys do, especially on the research and infrastructure side, so let’s just start with your research platform. So I mean, without a good research platform, you can’t really do anything. So, tell me about like building the research platform to be able to run your models, your risk management execution, let’s just start there.
John Krautsack:
Sure. So, first of all, the whole research platform that we’ve developed in house here and the front end trading platform was written by our two head of research Dave Polli, who’s been with the firm for over two decades and Pat O’Leary, who’s been with us about a decade and a half. So, they were integral parts of writing this software. Dave Polli has a degree in Electrical Engineering from IIT and Pat O’Leary… sorry about that. My phone-
Jason Buck:
No worries.
John Krautsack:
Pat O’Leary has a master’s degree in Financial Engineering. So, both these guys, brilliant guys, have been with us for a long time. And they wrote all this software, so all the software at EMC is all proprietary, written in-house. So, we get into the research, so we have like a very adaptive research process. And we believe this is why we’ve been successful over the last three decades. We constantly are evolving our strategies and our risk management through this process.
John Krautsack:
We currently employ machine learning which is basically a branch of artificial intelligence to help optimize our whole entire trading strategy. The machine learning allows for the risk management and systems to adapt to the changing market environments and that has been critical to us because systems
over time degrade and if you don’t reoptimize, you’re going to fall behind. So, the re-optimization process is basically what we’re doing is we’re looking at historical data and we’re optimizing that data to core parameters of systems, whether they be trend or momentum, and to our at-risk management overlay, which has its own core parameters that we continue to optimize to.
Jason Buck:
And part of that, that’s your forward walk optimization methodology is constantly reupping and making sure you’re adding and feeding clean data into the system to constantly be forward looking, or how do you think about that?
John Krautsack:
Yeah, that’s exactly right. So, we use a forward walk methodology along with, we used fitness metrics. Basically, what they are is a combination of metrics that we pair to each one of our systems, so they’re unique to just that particular system and we pair it to our risk management as well. And it basically allows us to define the contribution that we want each system to have. So, an example of a fitness metric, it’s a mathematical formula of a short-term system, say we’ll pair that to Sharpe ratio combined with an enhanced return numerator. And that allows the system to basically be more nimble, be more nimble. It’s looking at the risk constantly on that system.
John Krautsack:
Whereas our longer term system, we might pair it with a Sharpe or Sortino ratio combined with a return. And it really gives the roadmap for our systems during the optimization process of where they’re being directed. The longer term system is more comfortable with volatility, more volatility. And so, that’s, both systems could be very similar core logic, but during the optimization process, the look backs become very different between the two systems, the volatility thresholds, how you’re getting into these positions and a short-term look of how you’re getting out of the positions. So, the parameters, we might have 13 parameters that get us into trades along with getting us out of trades.
Jason Buck:
So, as you’re running the individual strategies and almost like they’re siloed pods, and you’re working on the fitness metrics for each strategy with the forward walk optimization, tell me how you’re done using genetic algorithms as an overlay to even improve on the fitness metrics.
John Krautsack:
Yeah, so those are some of our key ingredients. So, we have this machine learning process that incorporates the forward walk methodology. Then we combine each system with a fitness metric, its only unique to that one particular system. We then do genetic algorithms, through basically what genetic algorithms are doing. We’re populating these parameters with the best traits for us to successfully catch these trends. So, the best lookbacks and the best volatility filters, the genetic algorithms are processing through our cluster network of 200 quad core processors to come up with the best parameters per each system, per our risk management.
John Krautsack:
Our risk management is very unique. It’s, so each system has its own risk management associated with it, but we have a risk management overlay where we have several core parameters in it, looking at
trailing P&L, looking at open trade interest, open trade equity. And so, those are core parameters and we look at an acceleration past the certain threshold that we’ve optimized to. And if we get past that threshold, our risk management scales the entire portfolio across the board.
John Krautsack:
So, an example of that was like ’08. It was the perfect storm for trend following. We’re getting close to that certain environment right now in the classic program where our risk management has gotten past this optimized threshold and has scaled back our risk close to 50, 60% right now. That has been an incredible tool to our strategy. We get feedback constantly from institutional investors and our strategy that several times the next month is a big pullback in performance among the CTA community.
John Krautsack:
And we get feedback like what you guys do. “What are you doing, because you captured all this equity?” And it was really the reason for it. Everyone knows that CTAs have this volatile give back performance every once in a while. And we created this tool, but everything we do is informed by our research. So, we have to see the results before we implement something. And this has been, it’s been pretty good over the last, say 10 years.
Jason Buck:
And I want to go, we’ll do a deeper dive into a lot of details of your risk management in a little bit, but I want to go back. When I think about machine learning or AI, especially with AI, we think about broad AI, like artificial intelligence people are so scared of. But it seems like the best use cases narrow AI. If you have a really narrow set of parameters and like a close up system, you can train the AI to figure out the best risk metrics set or like your fitness metrics, or whether it’s the algorithm. But it’s almost like you have to train them on a very narrow niche set of problems that are hopefully well defined. Is that fair? Is that what you’re trying to find over time? Instead of having like, “Here, I’ll just throw a bunch of data and the AI will figure out how to trade for us.”
John Krautsack:
Yeah, we definitely do. We give the research core parameters, core objectives, and then through that machine learning, it’s looking for filling those parameters. And I understand what you’re referring to and we’ve taken that to research as well. We have not implemented something where we’ve looked at a million different, we’ve created a million different technical type strategies. And so, through our whole research engine, old strategies, new strategies, trying to combine different parameters from each one of those systems, whether they’re current systems or shelf systems, and trying to build a system that we didn’t even think about.
John Krautsack:
We try to do that in research and sometimes we come up with some improvements. But for the most part, your first analysis was correct. We basically give the roadmap to what we want these systems to look like. And then the JAK algorithms help us populate the parameters within that core logic.
Jason Buck:
You just said something that I think is probably one of the most difficult questions is like sometimes like you said, it will spit out a system or an idea that you want to come up with on your own. But then how
do you overlay, does it make rational or reasonable sense why this would work? And then so are you wrong with overlaying reason and rationality where maybe the AI and ML is finding something that like a human eye can’t see. So, there’s a trade-off there. Maybe you can’t see it or there’s a commensurate risk because it’s using back tests. The data you’re feeding it might be on a walk forward basis, there might be an exogenous event that you know as a human that could throw off that entire system.
John Krautsack:
Yeah, I agree with that. Yeah, it’s, I mean, there’s a fine line. And I think, one of the biggest hurdles as a systematic trend follower or a systematic strategy, is that you have to be very, very careful of your disciplines in your optimization. And what I mean by that is, you can’t just take all this historical data and build a system off of everything you know. You have to use out of sample data. You have to diversify your strategy. It’s just important not to overfit and it’s easy to do that. So, you have to have the discipline to not overfit strategies and build systems for one market that seems to work. That to me is a dangerous road to travel.
Jason Buck:
And then another piece, a core piece of your guys’ infrastructure and research is your customized execution. And I think this is probably one of the least talked about things in this space. It’s like, for example, if I’m listening to top traders unplugged and they have a lot of all of their interviews, and they have a lot of their audiences DIY. They want to create up their own trend following systems, they want to run their own trend following program. But I think the piece that’s missing, one of the primary pieces that people don’t talk about, is the customized execution.
Jason Buck:
And so, if you’re a larger firm, like you, that can run a lot of this data in-house, to have all of your systems and your entire team in-house is really working on, working those orders to get the best execution possible. There’s a lot of alpha in execution that an institution can get that an individual DIY trader can get. It’s that.
John Krautsack:
That’s true. I mean, once we put in our 600-sq. ft. server room, that houses a couple of a hundred servers, Quad Core processors, we were able to take on that approach to execution. So, we analyze tick data, so that’s a monster within itself. The processes are really maxing out their workloads through that cluster management software that was developed in-house as well. So, we’ve developed algorithms per each market, so depending on if a certain market gets triggered by a system, immediate immediately upon that trigger, we have our own custom algo that works at that particular order. So, it might start just putting in one last pegging the bid.
John Krautsack:
We have a whole bunch that are proprietary to us, the algos. There’s a lot off-the-shelf stuff out there, but we prefer to do it in-house. we know our how our systems are built and we know how we want to implement them through research. We know it’s more successful. You might want to if you’re trading Euro dollars or short Sterling or URI board, you could pretty much execute that stuff pretty fast because the size is there, the liquidity is there. Other markets like our 40% commodities, you have to be very flexible with how you’re going to implement these algos. And so, once again, we’re informed by our research and it has had an incredible effect on slippage.
John Krautsack:
We have positive slippage across a lot of our markets because of our execution algos. That in the past never happened. There was just, there was a lot of slippage in the ’80s and ’90s. You really couldn’t control. It was, we were working orders just like stop orders and you’d get ripped through those stops and you just. So, in our research we always add to our research slippage assumptions and commissions and everything, so we really know when the results come out this is what we’re looking to get out of the strategies.
Jason Buck:
That’s right. As I was saying, I think that individuals or small firms don’t realize that cost and like you said, if you have positive slippage, it’s even better working those orders. And then what you hinted at earlier, I also think is really difficult for the individual is that research platform that you built in-house with all of your team and your research team. Because people don’t realize, you need clean data dry to do any of this.
John Krautsack:
Absolutely.
Jason Buck:
And so, we’ll be picking through that data, making sure you build those datasets in-house is a tremendous advantage, because you don’t know what the assumptions that went into people even putting the dataset together. And I think people miss a lot of that a lot of times. So, I want to get into the strategy more of, especially with EMC classic like you alluded to earlier. You’re running four core systems, but they’re independent systems, but you’re applying them to all markets. Can you talk to me more about how that works?
John Krautsack:
Sure, so the four systems are risk weighted a quarter percent each and then within that, we have market weightings and the market weightings are, they’re all different. So, we don’t trade every market equal because correlations within a certain sector, like the long-term fixed income or currencies or metals or energies, there’s a lot of correlation within those sectors. So, we look at the correlations of those markets, we also look at the liquidity, and then we also look at our returns from our systematic trading strategies. All that comes together to decide what our market weightings are. So, once we have that all put together and built the overall classic strategy has a leverage, a starting out with leverage points of basically what we’d want that program to look like from a return deviation standpoint.
John Krautsack:
And then those market weightings, those system weightings do not change, so that we have a set risk per each market, per each sector and overall, the VAR of the entire portfolio. So, for an example is, if we had a market that was a fully weighted market, let’s say the Japanese government bonds. They’re sort of sitting out there unique to their own self, they would get like a full weighting. And what that would mean is, if every system, all four systems were positioned for that market, the max risk would be 0.4% versus a market like cotton. That might have a lower risk market weighting and then you have to divide that by the four systems to come up with a risk across the portfolio. Once again, it’s set risk. It’s part of what you have to do if you’re going to be a systematic manager, you have to set your risk levels across
the entire portfolio. And you do that by running the research, running the correlations and understanding the interchangeable nuances of the markets.
Jason Buck:
So, part of that is and this might be getting into the proprietary secret sauce, but when you’re putting on those positions is that’s how trend follows your CTAs is actually they manage a lot of the risk is in the initial position sizing. And historically that was ATR, Average True Range, which you’re basically setting the variance or volatility over a certain amount of lookback. But are using your proprietary models to figure out what is the actual initial position size of the trade? It sounds like you’re saying not only due to ATR variance volatility, but also correlations with maybe other trades you have on or correlations within each individual sector?
John Krautsack:
Yeah, so that part is not a moving part. So, when we set the market risk weightings, it’s based on the research that we’ve done over several decades of what are the correlations. How do the correlations look between bonds and 10 years high correlation, real-high correlation versus you’re going to see in the energy sector, well, natural gas doesn’t really have a high correlation to rent, crude oil, gas oil. Those are all pretty highly correlated, but natural gas sort of has been its own separate animal when it comes to correlation. So, we’re able to give it a higher risk weighting. And right now, that’s been a phenomenal thing that we’re able to put a lot more risk in that trade because it’s been exploding to the upside in a big way.
Jason Buck:
And so, I assume then, in general, like that individual instrument or market is going to have a trade size of anywhere from let’s just call it roughly 20 pips to 200 pips, but then when you go up to that sector basket, like you’re saying energy or metals, what’s the band of constraint across like energies or metals or sauce?
John Krautsack:
Yeah, so just the-
Jason Buck:
An idea, yeah.
John Krautsack:
The market weightings of that sector summed up and that basically is what your sector risk is.
Jason Buck:
But that’s not constrained at the sector level or it’s just whatever those individual classic-
John Krautsack:
If all four positions our systems were put on based on the market weight, that would be the constraint level. But what you have to take in consideration for and we do take in consideration, this is the expansion of volatility. So, you could put on today four trades with a max risk of 0.4, but if the volatility goes up 100%, that risk number goes up with it. So, that’s where our risk management overlay really
comes in. scaling back the portfolio. Right now, you see some good volatility in these commodity markets and even in stocks and fixed income. Because of the volatility increase and the profitability increase, we’re able to cut that whole portfolio back to pretty much like the original trading level that risk level that we took on.
Jason Buck:
And so, then we’re talking about market sector basket and then at the portfolio level, you’re saying you’re using like some var at the portfolio level. What’s the constraint at the portfolio level or the bands of constraints?
John Krautsack:
Band of constraints for our entire portfolio, it differs, obviously, with all our programs. Classic program can run as low as a VAR of three. And we’re talking uncorrelated here, not correlated VAR, so we can run a three or that thing can ramp up to 7%, so with a three standard deviation move. So, most of that time when we exceed our VAR, it has been on the upside.
Jason Buck:
And so, you touched on one of the eternally debatable questions in CTA trend, and that’s the idea of vol targeting, right? But I think there’s a lot more nuance to that question, right? So, as you open a trade and it stays in the lower vol environment as volatility expands. The question is, do you target the volatility open positions a lot or are you against it? Are you for it? I think yours is more nuanced is that you’re trimming positions. But then part of that question is, do you need trends to really move through expansion of volatility? What’s the correlation between volatility and strength of trend? And then now maybe you have deeper drawdowns of open equity. I asked like five questions in there, so I guess it’s choose [crosstalk 00:48:13].
John Krautsack:
That’s okay, because I always think when that question comes up, my partner in the business here, Brian Proctor, he always explains it to people when they ask that question is, “There’s good volatility and there’s bad volatility.” The good volatility is when we are seeing these expansion of trends and the bad volatility is watching markets just get chopped up in both directions. That typically is not good for a trend-following strategy. So, during those bad periods of time, our systems are very nimble and we’re getting in and out of trades and it’s, you know.
John Krautsack:
But we’re doing what we’re supposed to be doing, because we have a set risk, we make sure that we adhere to our systematic strategy, to the point where we have our research platform is separated from our front end trading platform. So, every day when we generate orders for all of our markets, globally, that trading platform has to match with the research platform. If it doesn’t, we can’t release any orders until we resolve why it hasn’t. So, you have to be trading the exact same size according to research trading. So, we really try to say 100% disciplined to a strategy. Probably, the only discretion in our strategy is really when it comes to the realm of spreading the markets from one contract to the other.
John Krautsack:
Where that really has an impact is in the commodity portion of our portfolio, you have to make assumptions in research when you’re running all this research and that assumption is when you spread is sort of just based on first notice day on liquidity that we’ve researched. But in trading, as we saw last year in crude oil, we were short crude oil and there was a massive squeeze in that front month to the point where prices went negative. I mean, we didn’t hold on to the position that long, even though we were short, and it was a squeeze, because there’s just, we don’t ever want to take delivery of a contract.
John Krautsack:
But so, we spread that contract a lot later and benefited from that spread versus what research assumed that we would be spreading the thing 20 days ago. So that part is the discretionary part on our trading team, just looking at what’s going on in those spreads. And what, are we long, are we short, what’s the spread doing, and how fast we should spread? We also took advantage of that energy, that energy crisis basically, by basically, we are able to then after that happen, spread out a couple months further out and we captured this incredible price spread between the front month that we are trading and the further out months of almost like a $10 crude oil move, we captured in just the spread, before it kind of calmed down and equaled out again. So, that’s about the only discretion we will use in our strategy and it’s just based off of our ability to spread when we think it is the best time.
Jason Buck:
Speaking of those spreads, do you ever think about like disaggregating the midterm structure and the carrier role yield you get from that term structure? Or do you think that just nets out in the wash across all your positions?
John Krautsack:
I think that kind of nets out, ultimately. Just basically, because we are long and short, it’s really like each instances is different most of the time.
Jason Buck:
That was probably a parameter just not a large parameter in part of your calculations.
John Krautsack:
Right, right.
Jason Buck:
Yeah. So, then going back to you have continual daily monitoring, but one of the things interesting you guys say is that you have an eternal degree of confidence of return stream measure to predict magnitude of outlier days. So, that’s unique. What do you mean by predicting the magnitude about outlier days on the position?
John Krautsack:
Yeah, so we have a live risk matrix on our front end platform. And basically, what it is doing is it’s calculating the risk across markets, across systems, across sectors, taking in consideration the expansion or decrease in volatility, so we have risk per market, risk per sector. And then we bring that all the way down to measure basically the VAR of the entire portfolio. So, that’s 100% live on demand, so we had a position. We exactly know what our whole portfolio’s risk is at, at any time.
Jason Buck:
Yeah. And it shows where my head’s at. Because actually, when I read that, I was thinking like that you were trying to predict the magnitude of outlier days on positive P&L, but you were actually predicting the magnitude of outliers on negative P&L and constraining the book that way. That what you’re saying.
John Krautsack:
Well, you can look at that both ways, because it’s the risk in the portfolio. So, if every market went in your direction by three standard deviations, then that VAR number would be the upside VAR or vice versa of everything. We’re so diversified that’s a really tough thing to happen. I mean, in ’08, we almost at every market had every single system positioned in it, just the perfect storm. And I believe during the ’08, ’09 period, we exceeded our VAR a couple of times, but it was to the upside. Just because the unique and negative correlation between or low correlation between their commodity part of our portfolio, and then the financial part of our portfolio, it would have to be the in perfect storm for us to exceed on the downside.
Jason Buck:
And then, just put a finer point on the covariance, it’s one thing to run your covariance matrix when you’re putting on the positions. But as we know, covariance changes through time, and it’s hard to get your arms around it. So, as it changes through time, are you constantly reupdating your models and that’s maybe why you trim positions if correlations are going closer to one across your asset classes? Or how do you deal with covariance matrix through the arrow of time?
John Krautsack:
Yeah, I mean, we’re going to set our market weightings well in advance based on our research going into the year, but we’ll constantly revisit that if correlations are starting to fall apart, historically. We continue to do all that research for originally how we built a portfolio and going forward. But once a year, we reoptimize our systematic strategy for each one of our systems and also our risk management overlay. We reoptimize to stay on top of it, so you’re not privy to what data is going to be, what the markets are going to do next year. So, every year, we add that New Year’s data and reoptimize the strategy.
John Krautsack:
And same with these correlations, where we look at the correlations, we reoptimize our algorithms for trading. That’s what we do. We constantly are adding new data, allowing all of our parameters to learn from the new market environment. And that’s how we avoid our systems and our strategies and our approach to correlations and everything. How we avoid those strategies to grading.
Jason Buck:
And I think we kind of maybe sidestepped the vol targeting question a little bit earlier. So, I want to come back to it in a way because you’re talking about when you’re looking at downside protection, upside potential, you guys use marginal utility measure to harvest profits and generate better risk adjusted returns with lower draw downs. So, I would, my assumption would mean that means that dimmer switch of harvesting that over time. It is not like maybe a light switch of turning on and off based on target volatility. So, tell me about how that dimmer switch works if it has that volatility expands and you want to start harvesting profits on different traits.
John Krautsack:
Yeah, so that, so basically, what this utility function is, is another metric that we use just for optimizing our risk management overlay. And layman’s terms, what that utility function is doing is when we’re optimizing a strategy, we’re trying to produce a product that has a specific return and specific volatility associated with it. And really, how we determine that in classic was really with our largest clients. We put this in front of them of, “This is the return, the comfortable return, the deviation. This is the expected drawdowns to this strategy.” And then once we know what we’re trying to achieve and what’s comfortable to the ultimate clients, I mean, we’ve had two institutional clients invested in EMC classics, one since 1989, the other since 1991.
John Krautsack:
So, classic, is what it is because the feedback from those institutional investors have they want classic to hit the cover off the ball when we’re in crisis mode or to catch these commodity moves. And they’re accepting of the volatility that we have to go through in order to achieve that return. So, getting back to your original question, which was, what was your original question here? I just forgot.
Jason Buck:
Just trimming those positions at dimmer switch. I think you answered [crosstalk 00:59:28]. Yeah.
John Krautsack:
Correct. Sorry about that, so using this utility function, basically what it’s doing is it’s rewarding a negative return during the research process more greatly than it rewards a positive return. So, it allows us to shape that distribution curve across the strategy. So, it’s basically avoiding the big pullbacks in the strategy. It’s like, “Would you rather have a 12% return with a 2% drawdown or would you rather have a 25% return with a 10% drawdown?” That’s what the utility function is helping us create basically within the portfolio.
Jason Buck:
You just hit-
John Krautsack:
[crosstalk 01:00:27] whether you’re scaling out of positions or scaling into positions, that’s a function of that risk management metric.
Jason Buck:
And you just hit on one of my favorite things why I’ve always been appreciated the CTA universe, you just talked about return versus drawdown instead of return to volatility like Sharpe ratio. And so, but part of that is it made me think of a few questions that we could wrap up on one is, I think, as the asset gathering business and large AUM of a lot of CTA managers has happened over the last few decades, understandably, part of that is they targeted a low portfolio vol because they said that’s what the clients want and so, they’re just giving them what they want.
Jason Buck:
But you guys are usually comfortable with running a higher ball. But let’s also be honest, running a portfolio target ball is a fictitious number, because you’re going to have bands around that depending
on what the market gives you. I’m just curious to how you guys think about running with your volatility or variants, your portfolio?
John Krautsack:
Yeah, I mean, that’s why we have several products, because classic isn’t for everyone. And so, we built some other new products that obviously lower the risk adjusted returns to that overall objective. And in one of our strategies, we basically added an element of long equities, long-fixed income, long gold to smooth out those choppy returns that we can get in the classic program, so that’s we’re not for everyone, that’s for certain. So, we have, like I said, long-term investors, who are willing to go through a little bit more volatility for a little bit more reward versus some investors who are really not comfortable.
John Krautsack:
We’re managing, we’re a subadvisor into a mutual fund structure and that program and performance is a lot more tamed. We’re aiming for single digit returns, single digit volatility, drawdowns no greater than 15%. So, we realize, managed futures, you have to really diversify your products that are out there. And in some cases, custom build those products.
Jason Buck:
Thinking about, I hinted at, there’s sometimes a marketing problem in the space and one of those was always trying to figure out what to call it. And for a while, the idea of Crisis Alpha kind of stock, right? And that was coming out of 2008 and they’re pitching Crisis Alpha across the space. And then when these shorter term drawdowns happened, and you didn’t see the commensurate response, because it’s uncorrelated, not specifically negatively correlated, they started to move away from the Crisis Alpha. But that was because of the difference between like we talked about earlier, long term and short term.
Jason Buck:
But as a medium term, without getting into details for compliance reasons, you guys have that have done really well in crisis period. So, are you comfortable sticking with that Crisis Alpha moniker?
John Krautsack:
Yeah, I think it’s not only if we’re comfortable, but it’s it really is if our investors are comfortable and the objective for us is just to maim discipline to our strategy and the objectives that we’re trying to shoot for each one of our strategies. And I think we’re a part of some decent portfolios, because we’re known as a firm that continually performs when there’s market disruptions. We’re always there for, we’re not changing our strategies just because things aren’t working out. We’re not eliminating markets. We’re not taking different types of risk.
John Krautsack:
It’s really a discipline process here and when we stick with it, it’s hard. It’s not easy during some of those periods of time, but when we’re supposed to deliver, at least to the clients that are in the classic program, that’s what they’re invested in EMC for is for our discipline to capture those moments when they truly need it the most. And so, that’s our goal is to build strategies that are fitting to each risk tolerance out there.
Jason Buck:
And then, thinking about this evolution or this constant iterative process of micro improvements, I’m just curious, on your guys’ take on what your research has led you to think or what you’re working on in the cryptocurrency space with like the Bitcoin and Ethereum futures. And what you guys think about the future of adding those, subtracting those, obviously, it depends on the client, it depends on the regulatory burden. But it’s just that it’s I think that’s on the tip of everybody’s tongue, so I’m just curious of your thoughts on the Bitcoin and Ethereum futures.
John Krautsack:
Yeah, we’ve definitely entertained it. We’ve also approached our clients about the possibility of adding something like that to our portfolio. Right now, I think the vol in those strategies is just so enormous that one thing I never brought up in our conversation about building our systems and the parameters in our systems is one of the parameters in every single one of our systems is a volatility filter. So, we’re looking at current volatility versus past volatility, and that volatility will screen us out of trades.
John Krautsack:
Well, from our standpoint, going through 2008, the one thing we can’t do is miss a good trend opportunity. We want to be there for all these opportunities and not be screened out. So, the optimization for each system, the thresholds are quite different and some are way more accepting of volatility, because we want to be in, in those trades. So, I mean, that’s like, that’s critically important to us. We’ve tried to get rid of them through our research process. Let’s just build a more robust system that doesn’t have to deal with it. They screen out so many bad trades during our research process that we need to keep that parameter in all of our systems.
John Krautsack:
So, I just recall back in ’08, ’09, one of our largest clients said to us, when all the commodities raised to new highs and then by probably through the first six months, and then every market reversed, one of the allocators in our clients, basically called us and said, “What are you guys doing? You still have risk on. I have a whole bunch of managers who have no risk on at all. Just clean the table up.” And so that’s one of the ingredients that we, we pride ourselves in is that we’re going to be there for these outlier moves. At least with one system, it might not be our total risk, but we’re going to make sure we catch those outlier moves.
Jason Buck:
Great. John, I want to thank you for coming on the podcast. If people want to learn more about EMC, it’s emccta.com. But thanks again for coming on. I appreciate your time.
John Krautsack:
Hey, thanks for having me. I appreciate it.
Taylor Pearson:
Thanks for listening. If you enjoyed today’s show, we’d appreciate it if you would share this show with friends and leave us your review on iTunes as it helps more listeners find the show and join our amazing community. To those of you who already shared or left a review, thank you very sincerely. It does mean a lot to us.
Taylor Pearson:
If you’d like more information about Mutiny fund, you can go to mutinyfund.com. For any thoughts on how we can improve the show or questions about anything we’ve talked about here on the podcast today, drop us a message via email. I’m Taylor@mutinyfund.com and Jason is Jason@mutinyfund.com or you can reach us on Twitter. I’m @TaylorPearson.me and Jason is @Jasonmutiny. To hear about new episodes or get our monthly newsletter with reading recommendations, sign up at mutinyfund.com/newsletter.