Shownotes
In his first podcast episode ever, Bill Eckhardt emphasizes that successful trading hinges more on risk control than on predicting market movements, a theme that resonates throughout his conversation with Moritz Siebert and Rob Sorrentino. The episode explores Eckhardt's journey from the renowned Turtle trading experiment to his current systematic trading strategies at Eckhardt Trading Company. The conversation provides insights into the evolution of trading strategies, the importance of maintaining emotional discipline, and the necessity of adapting to changing market conditions. The discussion also delves into the challenges of overfitting in trading models and the significant role that robust statistical methods play in managing risk. With anecdotes about the unique characteristics of traders and the importance of maintaining a diversified portfolio, this episode offers a fascinating look into the mind of a trading legend and the principles that guide his enduring success.
-----
50 YEARS OF TREND FOLLOWING BOOK AND BEHIND-THE-SCENES VIDEO FOR ACCREDITED INVESTORS - CLICK HERE
-----
Follow Niels on Twitter, LinkedIn, YouTube or via the TTU website.
IT’s TRUE ? – most CIO’s read 50+ books each year – get your FREE copy of the Ultimate Guide to the Best Investment Books ever written here.
And you can get a free copy of my latest book “Ten Reasons to Add Trend Following to Your Portfolio” here.
Learn more about the Trend Barometer here.
Send your questions to info@toptradersunplugged.com
And please share this episode with a like-minded friend and leave an honest Rating & Review on iTunes or Spotify so more people can discover the podcast.
Learn more about Eckhardt Trading Company here.
Follow Moritz on Linkedin.
Episode TimeStamps:
02:30 - Introduction
03:59 - Eckhardt's relationship to Richard Dennis
06:34 - Can trading be taught?
08:50 - The story behind the trading rules
12:32 - What characterises Eckhardt Trading?
13:44 - How did their perception of risk change over the years?
22:01 - Why they focus on short term trading
27:25 - How they use volatility in their trading
32:49 - Do they utilize other methods than trend following?
34:05 - Why do they avoid trade summary statistics?
40:26 - How they control overfitting and underfitting
45:37 - How they use "the gauntlet" to test systems
49:12 - How many markets do they trade and why?
53:57 - Why they don't trade single stocks
56:28 - Their view on replication strategies
01:05:23 - Evaluating the success of systems
01:08:02 - What are the most important things to avoid when designing trend following systems?
01:08:47 - The future of trend following
01:11:11 - Criticise ideas, not people
Copyright © 2024 – CMC AG – All Rights Reserved
----
PLUS: Whenever you're ready... here are 3 ways I can help you in your investment Journey:
1. eBooks that cover key topics that you need to know about
In my eBooks, I put together some key discoveries and things I have learnt during the more than 3 decades I have worked in the Trend Following industry, which I hope you will find useful. Click Here
2. Daily Trend Barometer and Market Score
One of the things I’m really proud of, is the fact that I have managed to published the Trend Barometer and Market Score each day for more than a decade...as these tools are really good at describing the environment for trend following managers as well as giving insights into the general positioning of a trend following strategy! Click Here
3. Other Resources that can help you
And if you are hungry for more useful resources from the trend following world...check out some precious resources that I have found over the years to be really valuable. Click Here
Transcripts
The structure of the market is that the money flows from the many to the few. So, you have to be among the few, not among the many.
If you do normal human reflexes, normal human reactions, then you'll almost always find yourself among the many and you'll be a loser. Trading doesn't feel good. If you're enjoying it too much, you're probably doing something wrong.
I tell people that to be a money maker is sort of negative, some, emotionally. You're going to have to suffer a certain amount to make money. You can't simply succeed by following your inclinations.
Intro:Imagine spending an hour with the world’s greatest traders. Imagine learning from their experiences, their successes, and their failures. Imagine no more.
Welcome to Top Traders Unplugged, the place where you can learn from the best hedge fund managers in the world. So you can take your manager, due diligence or investment career to the next level.
Before we begin today’s conversation, remember to keep two things in all the discussion we’ll have about investment performance is about the past and past performance does not guarantee or even infer anything about future performance. Also understand that there's a significant risk of financial loss with all investment strategies and you need to request and understand the specific risks from the investment manager about their products before you make investment decisions. Here's your host, veteran hedge fund manager Niels Kaastrup-Larsen.
Niels Kaastrup-Larsen:Welcome to another episode in the open interest series on Top Traders Unplugged, hosted by Moritz Seibert. In life as well as in trading, maintaining a spirit of curiosity and open mindedness is key, and this is precisely what the Open Interest series is all about.
Join Moritz as he engages in candid conversations with seasoned professionals from around the globe to uncover their insights, successes and failures, offering you a unique perspective on the investment landscape. So, with no further ado, please enjoy the conversation.
Moritz Seibert:Hey everyone, it's Moritz here and thank you for listening to episode number ten of the open interest series on Top Traders Unplugged. Today's episode is a real special one.
I'll be speaking with a systematic trading legend who has never before appeared on a podcast, namely Bill Eckhardt from the Eckhart Trading Company, or ETC in short. Also joining our conversation is the company's president and COO, Rob Sorrentino.
ading Experiment in the early:I’ll also ask Bill and Rob about ETCs extensive research lab and the methods they use to avoid not only the overfitting, but also the underfitting of systematic trading strategies. We’ll speak about why and how the human mind can be a traders worst enemy, as well as how the systematic trading space has changed over the years, and we’ll explain why Bill believes being risk averse is necessary for being successful in markets, and why ETC is so immensely focused on risk control.
Let me stop the intro here so that we can move over to what I think will be a really interesting and fascinating conversation.
Bill Eckhardt:The program itself was Rich's. Richard Dennis managed the turtle program and also managed the risk control of the turtle program. So those did not apply my ideas.
However, as I said before, I did develop many of the systems. So, in the program itself, I should only be considered kind of an advisor, a secondary figure. The running of the turtles was entirely Rich Dennis’ perview.
Moritz Seibert:But by the sound of it, you were the initiator and creator of the breakout-based trading rules.
Bill Eckhardt:Well, yes, that’s true, but there’s more to trading than simply having rules. You have risk management. You have certain attitudes toward profit, loss, and the like, and that was Riche’s. Rich controlled that part.
Rich Dennis and I had different ideas, especially about risk management. And, in the case of the turtles, we argued plenty, but it was his program. I always deferred.
Moritz Seibert:How did you, if I may ask, get to know Richard Dennis?
Bill Eckhardt:I knew him in high school.
Moritz Seibert:Oh, there you go.
Bill Eckhardt:And we were both actually interested in trading at that time, even though we were underage. We both worked, for a time, at the Mercantile exchange as runners, things like that.
So, we got to know each other in high school and became good friends, the sort of friends who would go out together, you know, four or five times a week. We were good friends for all that time, no question.
And then he was trading already as a professional. I was working on my PhD, having some disagreements with my advisor. And Rich said, well, why don't you take a leave and I come on the floor and trade? And I did. And I never went back. I never finished the PhD.
Moritz Seibert:How did this, I'm not sure if ‘argument’ is the right word, idea come about - like the trigger? Can trading be taught, or is it a natural talent that you must have?
Because something triggered the creation of the program. Was it that you and Richard couldn't agree on that point and there was only one way to resolve it which is to start the program, or how did that come about?
Bill Eckhardt:My point of view, if you talk to Rich Dennis, I think you might get a different story here. But from my point of view, we never really had an argument about whether trading can be taught. I mean, what I don't like about that legend, and it really is just the legend, is that it puts me in the position of making kind of an absurd assertion. I mean, what do we know of that can't be taught?
I mean, there's certainly aptitude, and, you know, some people are better at things than others, but almost everyone can be taught. Some people learn to on their own, and other people have mentors, but I can't think of another subject that can't be taught.
So, I never said trading couldn't be taught. That’s a distortion, as some of these stories tend to be. I did, Rich seemed to think that thinking about trading could be easily mechanized, that you could just use what he called weighted rules. And I thought, in fact, that it was much harder to mechanize what is needed for trading. But I never. I promise, I never claimed that trading couldn’t be taught.
I've learned so much about trading, both from my own research and from, you know, some other traders.
Moritz Seibert:Well, isn't that great that 40 years later, we get that straight?
Bill Eckhardt:Okay, thank you.
Moritz Seibert:For the first time, you've listened to it here.
But, I mean, interestingly, Bill, how did you… and you've mentioned you've created the trading rules, how did you come up with these rules? I mean, they're not falling from the skies
Bill Eckhardt:Well, my original plan, when I was very young, was that I would invent a theory of trade. You know, my background is in the history and philosophy of science and mathematics, and I knew a lot about how scientific theories were developed, and I felt that I could apply this expertise to trading.
Now, 50 years later, I mean, a half a century later, I don't think I really succeeded in inventing a complete theory of trading, but I have formalized certain parts of trading in a mathematically rigorous fashion. So, I believe there's been some success.
In other words, I have discovered some trading rules that I believe would make up part of a complete theory of trading, which I think is yet to be formulated.
Moritz Seibert:Do you think it can be formulated?
Bill Eckhardt:Well, yes. Science improves. If you look at things like the science of ballistics, initially, people thought that ballistic trajectories were circles. I mean, like from a cannonball, that it traced out a circle.
When they finally started developing it correctly, they assumed that the ball with the cannonball was just a point, which is false. And they assumed that it only had mass, which is false because it has more properties than mass. They didn't take into consideration wind resistance at all.
So, all of those things were incorrect. But that was the way the progress lay. Because you have to oversimplify a problem sometimes in the beginning. And I feel that there are certain, there's a certain similarity with trading. One can't take everything relevant into consideration. You have to break things down to the most important factors, even if there's some oversimplification involved, and then develop a theory. So that's the way I proceeded with my theory of futures trading.
I mean, in the beginning at least, there was some simplification, almost to the point that I was false. I was oversimplifying the problem. But as more research was done, we improved.
I mean, eventually they began to learn how to take wind resistance, how to take the size and the shape of the cannonball into consideration. Ballistics improved. And I think you could tell a similar story about my trading techniques.
Moritz Seibert:Yeah, that's a good segue into the overfitting and underfitting discussion or question that we'll speak about a little bit later, and the work that you're doing inside your research lab.
Rob, you're the company's President and COO. If I were to ask you, give us a quick overview of the Eckhart trading company, ETC, in short, what would you say? You are a short-term trading firm. What sets you apart?
Rob Sorrentino:Well, I think I would say we're a research lab first and foremost, because we focus a lot on that. And out of that we're fortunate that we have some really unique characteristics in trading that have been very successful and copied by many people and created by Bill.
We're always striving to deliver those to our clients as best that we can. And obviously, given the dynamics of how the markets have changed, in a sense that they're global, 24/7, all different types of participants, it keeps us on our toes, particularly in research, to try to stay ahead of the curve. So, pretty much, I think foremost, we're a research lab that delivers trading systems. We do it for ourselves and have been doing it for a long time, quite successfully.
Moritz Seibert:Bill, when the turtle program ended, when was that, ‘86 ‘87, something like that?
Bill Eckhardt:Yes.
Moritz Seibert: And then you started, ETC in: Bill Eckhardt:Well, I should just mention that the risk management of the turtle systems was all owed to Rich Dennis, not to me. Rich Dennis and I, throughout the turtle years, had disagreements about risk, and I don't think it's a secret that between the two of us, I was the more risk averse trader.
My own attitude, or at least that was developed over the years, is that whereas most traders, of any strength, focus on profitability, in other words, kind of predicting where the market's going to go. I feel that we can't be very successful about predicting where the market's going to go. It's too close to random. It isn't random. It isn't a random walk, but it's too close for simple predictions to work.
Instead, you have to focus on risk control, because that's something. You can know your risk, you can change your risk, you can control your risk. You can't really know where the market's going or control that.
What you have to do is not predict where the market's going. You have to know what you're going to do when the market goes anywhere. The motif of my trading has always been risk control first.
Now, of course, you need profitability, but to realize that profitability, you need correct risk control. And that's why I use utility theory. This is mathematical utility theory, which has been developed actually for more than 100 years by some important mathematicians. And it's a part of what is called decision theory, and it has to do with how to make decisions everywhere in life, not just financial decisions, but all decisions.
What utility theory says about this is that every dollar, every new dollar that you make is a little smaller than the last one. And more importantly, every dollar that you lose is a little bigger than the last one.
I mean, one can see this is obviously true. $10 means a lot less to a millionaire than to somebody whose net worth is $1,000. It's just clear that the utility of a given amount of money is different depending on one’s situation.
So, utility theory is a mathematical way of taking all these things into consideration. And utility theory introduces the notion of risk aversion not just as some kind of vague psychological concept, but as a rigorous mathematical concept which can be applied to a large variety of situations.
Now, there are an infinite number of possible utility functions, but I have specialized, I've done research, discovered which ones specifically apply to commodity trading. And furthermore, I've discovered a small class of utility functions that work for different clients who are in different phases in their career and have different amounts of money. So, these are the utility functions that we use.
Now, I should mention, as far as I know, all other traders that I know of have this approach. First, they try to develop something, whether systematic or intuitive or fundamental. But first they try to develop something that makes money, that shows a profit. Inevitably that will be too risky. So, they graft risk control, almost as an afterthought, on their system, on their profitability system.
I've compared this to that horror movie, the island of Doctor Moreau, where they'll graft the head of a lion on a giraffe or some crazy thing. That is definitely not what we do at ETC.
At ETC we take expected utility, It enters in at the ground floor when we're brainstorming - really just thinking, throwing ideas out there. What we look at right from the start is how these ideas affect expected utility.
Now, another thing I could mention that might be informative is you have, like I said, there are an infinite number of utility functions. Some of them are risk neutral, and some of them are actually risk loving. Now, I consider all those utility functions, the neutral ones and the risk loving ones, to be demented. These are utility functions for somebody who wants to keep betting the ranch. And eventually, I think they're financially suicidal.
Also, as a technical matter, I've been able to prove that with risk loving utility functions, that it's possible to create bets that have an infinite expected value. Here's a bet where it says your expected value for taking this bet is infinite, but in fact, the bet isn't worth $0.50.
So, to me, this is a ridiculous, absurd risk-loving utility function. I mean, it just shows that it's crazy. But I mean, I have also seen, that's a very theoretical thing. It's based on the old St. Petersburg paradox, which is a couple hundred years old.
But I have also seen practical cases where people who loved risk, who had a positive attitude toward risk, I've seen them lose all their money time and time again. One has to be risk averse. That's the only thing that can work in this business. I'm not saying that's the right attitude in life everywhere, but in futures trading, risk aversion is primary.
Moritz Seibert:Yes, we're dealing with leverage. We're facing the risk of ruin if we're over betting. These are all very important points. Now, you're trading short-term.
Would you say that this is also a form of your risk management, given that you have shorter term systems, and therefore, in expectation, smaller drawdowns? You get in and out of trades quicker. You realize losses (as inconvenient as they are) quicker. Is this one form of expressing the risk aversion that you have?
Bill Eckhardt:Well, yes, I feel that we're short-term only in terms of what other traders, who are very long-term, say. I mean, we have short-term, we have medium, we have long-term systems, but we have trades that have a four or five day average length. We have trades that have a 50 or 60 day average length. Now, maybe for some people, maybe that’s considered short-term, but we actually think we’re sort of covering the short, medium and long-term perspectives.
Moritz Seibert:Some of, or maybe many of your turtle students and systematic, diversified, trend following traders who used to trade short-term. They have moved over to longer-term hold periods because of the alpha decay that we could observe in the shorter-term systems.
Now, the longer-term systems, 200 day timeframes, things like that, to the present day, they continue to work. They go up and down, and they're robust with drawdowns. But I guess you have made a conscious decision to not trade these systems and also to no longer use the indicators that you, Bill, have developed in probably the late seventies, early eighties based on breakouts and these types of techniques. You have moved away from the, yeah, the short-term systems. You're not trading the long-term timeframes, and you're no longer using breakouts.
Please explain.
Bill Eckhardt:Well, yes, the turtle systems were largely, and the turtle risk control were largely disclosed. I mean, we had one turtle who more or less was dishonest with us and took our Quartz Manual, which was proprietary, and published it under his own name. So, hundreds, thousands, maybe hundreds of thousands of people had read about this either on the Internet or in some other way. So, this has all been very disclosed.
Now, this doesn't mean that such techniques are absolutely useless, but they've been heavily degraded since we let the turtles go, and all this information got revealed. So, even while the turtles were still in operation, under our management or under Rich Dennis's management, I was already developing systems that were as different from the turtle systems as I could make them and still be trend following, in keeping with my principles. So, I felt it was a necessary development to move away from the turtle trading systems.
Some people even went so far as to say that because of the turtle systems, trend following was dead. In fact, I've heard about the death of trend following pretty much every few years throughout my career. And I'll have to say again, like Mark Twain said, the death of trend following has been highly exaggerated.
I still believe, although there are other ways than trend following to deal with the market, and we do a certain small amount of that, not counter trend, but just systems that aren't trend following. The majority of our work is still devoted to following the trend.
And because there's bigger competition, better competition, because more people are learning more about technical analysis, one has to keep researching to stay ahead of it. I used to compare it to the Red Queen in Alice of Wonderland, where in her country you had to run as fast as you could just to stay where you are. And I feel that that's the situation a commodity trader is in. He has to improve as fast as he can just to hold his own.
Moritz Seibert:Rob, when you and I spoke during our pre call, you mentioned a volatility based or volatility driven trading signal, which you are mainly using. I presume this is not a volatility breakout technique. It is something else.
Would you mind explaining to us a bit more what that is or how it roughly works? If you can. You don't have to give away any secrets, but it seems to be a different technique that probably not a lot of people have come across.
Bill Eckhardt:Well, if I could just talk about this, our trading is volatility based. I mean, volatility enters into our trading system in several different ways, in about five or six different ways. I mean, you can use volatility to size your trades. Something that we've always done. The turtles even did that. We even had the turtles do that. But you can use volatility as a metric to measure price change. And there are really various other ways in which volatility can be used. On the other hand, we still have a price-based system. It's based only on price. So, I wouldn't characterize our systems as volatility based or as price-based. They're based firmly on both.
Moritz Seibert:So, you're using volatility in several different steps.
Let me just ask you, are you normalizing all the markets that you're trading to a same level of target volatility, and then you run your signals on these now neutralized equivalent instruments?
Bill Eckhardt:Well, we're not targeting volatility as that's used in the business, because in a lot of situations, increased volatility often helps us. We don't want to restrain it. Nor on the other hand, when our volatility is too low, do we want to juice it up. We prefer to let the systems run as they were intended.
So, we don’t target volatility, volatility in the whole portfolio, but we’re very cognizant of the volatility of the individual futures. And it affects our trading situation. It both affects trades, and even more so, it affects risk control, because increasing volatility, other things being equal, means increased risk.
Moritz Seibert:And greater fluctuation.
When I speak with my friends and people from the industry, I always say that to me, personally, the most important aspect of risk control is the initial position, sizing and keeping losses small. You don’t want to trade too big. You don’t want to over trade.
Now, it then becomes a very separate question to what you do with a position and with a trade once it is in existence. Do you maintain that position as a constant from entry to exit? Or do you have that position dynamically respond to changing correlations, to changing volatilities, to changing whatever value at risk calculations? Different people do different things, but then it becomes a trading system that responds to all sorts of things, if you see what I mean.
Bill Eckhardt:We try to keep our original trading size in a given position. We try to keep that fairly constant. I mean, there are some times where we make small adjustments, so it's not a pure case. But I would say that, roughly speaking, we liquidate trade with the same position as we initiated it. That's true enough. Now, there are situations where you have to make small adjustments for technical reasons.
Now, we've tested adjusting to current volatility and other things, but since volatility often increases with a trend, we have found that, and if we were readjusting to volatility, we would be decreasing our position with the trend. And we've found that to be counterproductive.
Moritz Seibert:Yeah, it is, depending on how things go, essentially executing a hidden mean reversion trade against the direction of the trend. That is not always true by definition, but it is true in my experience, more often than not.
And in combination with us having a stop loss level anyways, which means controlled risk and therefore a synthetic long optionality profile if you will, were essentially long volatility or volatility seeking to a certain extent.
Now, I had something in my head, I don’t want to forget it. You said Bill, and maybe Rob, you, as well, that most of your trading is based on following trends, and it’s the combination of volatility and price. In that there’s a little bit (and I think you, Bill, said it's not countertrend), but there's a little bit of a different thing that you're doing - mean reversion, carry, I don't know what it is. Is it a mean reversion type of system that you're adding to the trend systems which you run?
Bill Eckhardt:No, it's not. I wouldn't call it mean reversion. You know, it's volatility that mean reverts. Price doesn't mean revert, at least futures prices.
So, I would just say that we have some methods that aren’t trend following, but that aren’t countertrend either. They simply ignore the trend and are based on other things. But I also want to mention that this is a relatively small part of our overall portfolio, maybe 20%. 80% of our systematic work is still trend following.
Moritz Seibert:In an article about your firm in the Hedge Fund Journal, I read about the E-Score. I want to ask you about the E-Score in a second.
And I also read one thing which I found interesting, which is that you avoid (and I wasn't sure if I understood this correctly), you avoid the usage of summary statistics or trade summary statistics.
Now to me, and I may get that wrong, these are things like average loss per trade, number of winning trades, total number of trades, all these type of things. And I think personally, to me, they are important. So, if they're not important to you, why not?
Bill Eckhardt:Okay, I believe that my remark about summary statistics, it was easy to misunderstand. And I'll try to be a little more explicit. You definitely do use summary statistics to grade and assess performance.
I mean, the kind of things you mentioned, like average profit, average loss, success rate, those are all summary statistics which assess performance. Now, I was thinking more in terms of summary statistics for trading signals, not to assess performance, but for the trading system itself.
And let me give an example. Let's say you're looking at the last 20 days just to really pull a number out. The last 20 closes. If you use a moving average, a simple moving average, you take those 20 numbers and boil them down to one. And now you only have one number. And that's what I mean by a summary statistic.
Now to change, to give another example, suppose instead of using a moving average, you just use the high of the last 20 days, the high close, and the low close of the last 20 days. Now you have two numbers.
Well, now you can tell if the market at the current price is closer to the highs or at a high breakout, or closer to a breakout; is closer to the lows, or at a low breakout, or somewhere in the middle. You have so much more information just from going from one number to two, putting in a little structure. In other words, instead of boiling all the structure down into one number, you're keeping some of the structure.
Now you can go further than keeping two numbers but of course, if you were to try to keep all 20 numbers, all the structure involved in the last 20 closes, it would be combinatorially prohibitive. You'd have millions of cases, maybe billions. You wouldn't have enough data to even have a single sample for a lot of your cases.
So, what you have to do is you have to find out, you have to eliminate a lot of the structure, but you have to keep enough of it. You have to keep what's important. And of course, that's one of your… What is important is, of course, one of your major research problems.
My talk about summary statistics was mostly a diatribe against using moving averages. Now I want to mention that we use moving averages to estimate volatility. Volatility is something, at least arguably, is something that statisticians call a moving average process. So, it's appropriate to estimate volatility using moving averages.
Price is absolutely not a moving average process. It's not stationary. It has none of the qualities of a moving average price.
So, the use of moving averages, it's a funny history here. Moving averages definitely have a use in certain financial time series, for instance, in economics. But I feel that they don't actually, they aren't actually useful in trading. And in fact, if you look at the way traders use them, you know, moving average crossovers and things, there's really no theoretical support for that.
You know, traders have a tendency to use ideas in a bizarre way. I'll give a real bizarre example. I've seen trading compendia, which say that you can trade gold based on the sun and trade silver based on the moon. Now, I hope everybody listening thinks that's as crazy as I do. I've seen trading systems based on the electron levels in atoms. These things are just irrelevant.
No time series analyst, no professional time series analyst in economics or another time series would say that crossing a moving average is significant. But that's what futures traders have clung to, this idea.
So, I would say that, in general, moving averages for trading for price are counterproductive. Moving averages for volatility, they're an important tool.
Moritz Seibert:That brings us, Rob and Bill, to the big research lab that you guys run. I guess that's the work that's happening there is figuring out when to use moving averages or when not to use them and how to use them.
I mentioned the E-Score, and Bill, I'd like to speak with you, and with you, Rob, about not only the avoidance of overfitting, but also the risk of potentially underfitting and not making use of the full opportunity set that the markets may present to us.
Bill Eckhardt:First of all, there are a lot of traders who talk about curve fitting. Curve fitting is a word from nonlinear regression where you have data and you try to fit a curve to the data. Thats not what you’re doing when you’re trading. I don't really know anyone who does that. So, curve fitting is simply the wrong term.
I prefer the term overfitting, as obviously you do also. And an important thing about the idea of overfitting is that it does imply that you can underfit. One's purpose is to try to fit to the permanent features of the time series, whereas you want to ignore the transitory features.
Now, this can't be done… This can largely be done, but it can't be done 100%. If you fit, and if you take fitting seriously, you're going to overfit a certain amount.
We have a bunch of techniques that have all been developed in house that, as far as I know, no one else uses to control overfitting. In other words, it's one of our major focuses to overfit as little as possible.
And as I said, we have designed… Well, I've designed techniques in house that are very accurate in estimating how much we're overfit. And I would say that over the years, we've really been using these techniques and improving them. We've really decreased the extent to which we've overfit.
Now, of course, if you're afraid of overfitting, one way is to underfit. But you see, that's really not a good procedure either, because you're not making use of all the information that the market… of all the worthwhile information the market's giving you. So, overfitting is a real problem. I know specifically of traders, good traders, who've lost all their money because they overfit.
Okay, now the E-Score, for a long time, for decades and decades, we've used different performance measures because there are different ways of assessing performance. Eventually, we decided that these had to be combined into something that back, oh, 20, 30 years ago, we called a supervalue. Other people, including some of the turtles, have adopted this terminology. So, we stopped using it because the fact is that we kept developing this combination of performance indicators, and we now call that the E-Score. It's a measure of performance. And there are various ways that a system can appear good, and that can actually be a false impression.
For instance, I'll just give one example. There are dozens. A system might make too much of its money over too short a period of time. Well, we have elements of the E-Score which will test for that and which will discount a system that makes all its money, let’s say, in two years out of the last 20. But there are a lot of other ways that performance measures can go wrong and give you misleading results. Our E-Score represents our best effort at finding different measures of performance so that different possible ways of failing can be avoided.
Moritz Seibert:Rob, I think it was in this context that you mentioned the gauntlet. Do I remember that correctly?
Rob Sorrentino:Well, Bill has spent as much time tearing systems apart as we do putting them together. So, the formal process that they've had here for many, many years was aptly named the gauntlet. And it's just a barrage of testing mechanisms that Bill created to try to tear a system apart. And it actually goes back to what he was just talking about.
We very much rely on that E-Score and looking at how it can help, particularly an overall blend when you're putting systems together. But the gauntlet is something we use. I mean, we have people that call us and want to have us run their system through the gauntlet just to see what the testing results would be. And we've had some younger students that we work with at the University of Chicago that have ideas, and we would run it through and then give them the results, because there's a lot of value in them knowing where the failures are in the system. It's just as important as knowing why it's working. So, for us, it's fundamental to everything that we do here, and it allows to take a lot of the emotion out of it. We strive for that.
I think anything we've all learned from Bill is to try to be as scientific as possible and just eliminate the emotion from the decision making. Because you can, you could have systems that they might have a two, three, four month, really strong run, and then they just go flat or degrade to start losing, and you get a pretty good, strong high when they're working, and then all of a sudden, they stop. So, all of those are small features of the whole DNA of the firm and how we use it.
But Bill could probably comment on where he came up with the name. I thought it was genius when he came up with it, because it really explains what we do when we do the testing mechanism.
Bill Eckhardt:One of the things I'd like to mention is that the gauntlet not only tests systems, you know, if there's a problem, various problems with the system itself, but it actually tests the program, too. It can find very subtle glitches in the program that otherwise would be nearly impossible to find.
Now, you know, I'm not really a secretive person. I would really like, for instance, to publish some of the gauntlet ideas just to show how clever I am, I suppose. But this is an industry where we have to keep secrets.
So, I have published scientific work in philosophy of physics, in mathematics and in finance, but nothing specifically. I haven't really published anything specifically about futures trading, as much as I would like to. But the way the world is, one has to keep these things secret.
Moritz Seibert:I'd like to speak with both of you about markets and the number of markets, but also the type of markets which firms trade. Let me be more specific. “Alternative markets” is actually a term that I don't particularly like, because to me, markets are markets. You know, it's corn, and it's soybeans, and it's an equity index. Whether power, or freight contracts or an exotic currency is alternative, I think is a subjective definition. I think to the people that trade coal and power, they're probably standard markets. It's just not particularly standard to the traditional CTA industry.
So, the first question is, and some people say these markets have better trending properties and they're easier to trade and they have higher risk adjusted returns when you put them into a trend following system. What would you say to that? Do you trade any of these markets? Do you find them interesting?
And then second (a two-part question), second question is the number of markets. I've been saying, and I'm not the only one, that like, we're in the camp of maximizing diversification, because it is something that we can essentially do for free. And not essentially (let me take that back) we can do it for free. The computer does the work. You know, diversification benefits are available for us, and we can harvest them. And the more markets we trade, the more independent bets we can place, the better we can crystallize our edge over time.
And the fewer number of markets we trade, the greater the impact of noise in any given sample or in any given run. But obviously it has pros and cons. There are periods where the portfolios that trade a smaller number of markets vastly outperform the portfolios that trade more markets simply because of the fact that they have more size on, and they're positioned differently, and the trends that currently work, and then the opposite can be true as well. So let me stop here. I think you know where my questions are going. I'd like to hear your views on that.
Bill Eckhardt:Yeah, we trade about 70 markets so potentially, I mean, we don't always have positions in all of them. Sometimes systems can be long, they can be short, they can be out of the market. So, we do really believe in diversification. I agree with everything you said about it.
Diversification was the greatest financial idea of the 20th century. And despite the power of that idea and its usefulness, I still think a lot of traders just like to jump on the latest bandwagon, the latest few things that are hot and avoid everything else. They are anti-diversifying.
The reasons we don’t trade every market that we might, you’re talking about some of these so-called alternative markets, is because we measure the liquidity and the trade slippage for all markets that we trade or that we’re considering trading. If there’s too much slippage or if there’s not enough liquidity to do the sort of volume that we need to do, then we avoid that market.
The thing that you mentioned, incidentally, that some of these so-called alternative markets seem to trade better, I would emphasize the word ‘seem’ to trade better. You have more slippage, so it's actually harder to make the trades.
So, on paper, the markets going to seem to be more profitable, unless you're really compensating for the troubles you're going to have initiating and especially for the troubles you're going to have liquidating when you need to get out in a hurry.
So, I would say that on paper, these alternatives, markets might appear better, but in practice, I think they’re actually worse, at least for the kind of trading we do.
Moritz Seibert:Connected to that, and even getting to a greater extent of diversification, is the trading of single stocks and following trends in single name equities. So, you’re breaking up the equity baskets, the S&P 500, for example, into its constituent parts. And you’re now following trends on the individual names. A, do you do that? And B, if not, why not?
Bill Eckhardt:Well, I don’t claim any expertise about stocks. I’ve studied them, certainly. Superficially, stocks are like futures. They’re price series, they’re time series, they go up and they go down. But they’re really profoundly different. When was the last time you seen a commodity split, or have a commodity issue a dividend? Or for that matter, the stock market is dominated by people who have long positions. It's dominated by the longs, whereas in the futures market, there is exactly one long for every short and vice versa, it's completely balanced. So, they are different processes.
Now, individual stocks, I mean, you can theoretically get a lot of diversification there, but stocks don't really have the kind of robust trend component that futures tend to have. In other words, our systems don't apply very well to stocks. And for that reason, we simply specialize in futures.
Stock markets are simply a different phenomenon and require entirely different rules. We would trade stocks. If I felt that I had some kind of relative advantage in stock trading, I don't feel I have. I mean, I'd be competing with people who've been studying stocks all their life. I've been studying futures all my life, and I feel that that's where my relative advantage lies. And so, we don't trade stocks.
Moritz Seibert:Another thing that we have seen in the recent months, years, is the emergence of replicating trend replicating strategies in ETF wrappers. And there's bottom-up replication, top-down replication. There's a relatively large fund available in the United States. There's not that much going on here in Europe. Maybe not yet. What's your view of these, Bill and Rob, like replicating strategies?
Bill Eckhardt:I'm not too familiar with the term replicating. The way you're using it. Could you explain it a little?
Moritz Seibert:Yeah.
So, there's, for instance, a method of using a top-down replication, whereby you are trying to replicate, for instance, through a linear regression, looking back a number of days, the returns of the SocGen CTA index.
So, you form that. You try to get an informative view. You define a number or a set of markets, say,10 to 20 markets, and you're looking to find the combination of these markets to create the best fit for the past N day of the SocGen CTA return.
So, you're not actually trading a breakout system or any of these rules. You're using the index returns as the input. You kick off an optimization process, and you find a fitting portfolio, which you then hold for a number of days.
Bill Eckhardt:It sounds like an invitation to really overfit.
Here's a fundamental thing, and that is that the distribution, the probability distribution of price change, and also of trades in a futures trading system, are not normally distributed. In fact, they're radically non-normal.
And this means that normal distribution theory (which is mathematically, by far, the most tractable), and normal distribution theory (which is the theory that you can get the most information out of your data), if you can assume the data is distributed normally, simply does nothing apply to futures trading. The futures are so non-normal, and they're so far from normality in their behavior.
I mean, there are a lot of theoreticians, I would mention the inventor of fractals, Benoit Mandelbrot, who say that the futures markets don't even have a variance. There's no variance. So, they don't even have a standard deviation.
So, all these people, for instance, who are computing the standard deviation of something in commodity futures are really dealing with nonsense. The kinds of statistics that can squeeze (these are normal statistics), the kinds of statistics that can squeeze meaning out of small samples 30, 40, 50, 100 have no place in futures. I mean, I wish they did. It would make it easier for everyone.
But the fact is that you need to avoid normal techniques, and you need to use what statisticians call robust or distribution free statistics, which are statistics that make no assumptions about the distribution from which are sampled. Now, these statistics, the drawback of such robust statistics is that they're kind of data hogs.
You need 10, 20, 30 times as much data to make a reliable inference. So that's a problem. But I would say that if you use normal statistics, you simply get the wrong answer. So that's a bigger problem.
So, if you're fitting to the last couple of years, let's say, if you're fitting to the last two or three years, you're not going to have enough data to use robust statistics. You’re going to have to use normal statistics, and that’s going to lead to problems.
I would mention many decades ago that famous company, Long-Term Capital Management, that had some of the greatest academics, including Myron Scholes, on their board from the black Scholes model. This company busted out spectacularly, and it really was because they were using normal distribution theory and its correlatives.
Like you can't use normal distribution theory, you can't use least squares correlation. In fact, anything where you're squaring data will not work in futures. The tails of the future distribution are too heavy.
We use only robust systems, which means that we need a lot of data, which means that we need 10 to 20 years of data. We can't draw any valid conclusions out of two or three years of data. Thats just overfitting.
Rob Sorrentino:I think that on the replication question, that’s probably not going to have a good ending. I’ll go on record to say that.
Moritz Seibert:The book that you’ve mentioned and the author you’ve mentioned, Bill, Benoit Mandelbrot, the book that made him famous is The Fractal Geometry of Nature.
And then later on he wrote a book called The Misbehavior of Markets, which I highly recommend, where he shows exactly, I think, with the example of cotton prices, that the assumption of normality in return distribution is nonexistent.
In fact, it's actually unclear what type of distribution it is. It is definitely tailed in both directions. It is prone to jumps. It is more like a Cauchy distribution, where you have not even a defined mean, the mean changes or its undefinable.
Bill Eckhardt:It’s probably not as bad as a Cauchy distribution, but I don’t think it has a variance. Mandelbrot suggests that it could be what’s called a stable Parisian distribution.
Moritz Seibert:The point I wanted to make, and this is interesting that you’ve just mentioned it, things like the average true range, which is not a standard deviation, but many people would calculate volatility as a function of standard deviation under the assumption of normal distribution. There you go again, you're back into normal statistics, with all its drawbacks.
And Rob, maybe to your point, whether it's going to have a bad ending, I cannot forecast that. What I dislike about these regression based replication techniques is simply the fact that I think we, as traders, owe a lot to the fact that we keep losses small and that we very unemotionally get out of trades if they start working into our core capital and they consume too much of our money. We throw the trade away, because we can try another time in a different market.
So, these predefined rules to exit positions, be that based on an initial stop or later on through a certain exit mechanism, they're so important because keeping these losses small is what keeps you in business.
If you don't have that risk protection, like you come up with a portfolio based on whatever statistical technique that you're using, and you're implementing that portfolio without that downside protection, without a known point at which you would definitely throw the trade away, that means that you are at risk of discontinuity and gap events and all these type of things which, you know, may not happen for a long time, up to the point where they all happen, or it does happen, and then it's too late.
Bill Eckhardt:Yes, I agree with that wholeheartedly. And I've seen evidence of non-normality in every futures market, not only in cotton.
The market distributions have heavy tails. So, in particular, the left hand tail, the one that represents losses, is very heavy. The technical name for this, in statistics, is leptokurtosis.
Now, all markets suffer from this leptokurtosis, and to some extent, all trading systems do too. So, the trading rules have to intervene to keep those left tails under control. And that's, of course, one of the major uses of risk management.
There are also, I should mention that there are systems, and a lot of people use them, that have very high success rates. They make money 70%, 80%, 90% of the time, but then they're subject to large freakish losses.
I mean selling cheap options is an example of such a strategy. Now we, on principle, avoid all such strategies. We like a trade success rate that’s around 40%. In other words, we have more losers than winners. But, of course, the winners have to be significantly larger than the losers. If I’m looking at a new system and it has a success rate of 70% or 80%, that's a prima facie reason to reject it. And in fact, if you examine such systems over longer periods of time, they periodically bust it out.
Moritz Seibert:These systems tend to be systems that exploit and depend on what you've just mentioned, as leptokurtosis, which is that most of the observations happen around the means. So, you have a greater sample, statistical sample of oscillation around the means where you can exploit that noise.
But obviously at some point these tail events materialize, and you have the outlier price move and that’s where your losses live and then they become very large because you’re unprotected at that point.
If you, Bill, with all the experience that you have, if you were to design a trend following trading system from scratch today, what would you say? What is the most important thing to avoid
?
Bill Eckhardt:Put ten times as much emphasis on risk control as you do on trying to predict where the markets going. You almost never know better than about 51% or 52% which way the market's going. I mean it's the natural human way to think about it, but it's not the correct way for success. Think about risk control. Your profits will take care of themselves.
Moritz Seibert:What do you think of the future for trend following?
Do you think it's a mainstay strategy with longevity and it's going to be around because it exploits human misbehavior and biases and we can just keep on doing it?
Bill Eckhardt:I mean it does exploit human biases and misbehaviors. I think a lot of traders have had the experience of, let's say, being at a party and talking, having a friendly conversation with somebody who's really never traded. And what I hear most often is they'll say, well if you have a loss, you don't have to get out, right?
Now, any experienced trader knows that that's wrong. You have to get out of your losses, you have to let your profits run. But here are people who've never thought about it and their first impression, right out of the box, is dead wrong. So, there's something about normal human attitudes that don't work in trading, so you have to somehow overcome those normal human attitudes.
The structure of the market is that the money flows from the many to the few. So, you have to be among the few, not among the many. If you do normal human reflexes, normal human reactions, then you almost always find yourself among the many and you'll be a loser. Trading doesn't feel good. If you're enjoying it too much, you're probably doing something wrong.
I tell people that to be a money maker is sort of negative some, emotionally. You're going to have to suffer a certain amount to make money. You can't simply succeed by following your inclinations.
Moritz Seibert:I think that is a big, big part of the equation is this pain arbitrage, of sticking to your system during bad times when the going gets tough and nobody enjoys it. That's the pain you endure. But you can make it out the other end because you've kept your losses small.
Now I guess we have to end it with a turtle related question. I was just thinking about how was it like, you have these applicants, men and women coming into a room. You invited them, you interviewed them, and then they're sitting there or standing there. You meet them for the first time, and you spoke to all of them, I guess, Bill. Was there one where you thought, or a couple where you thought, these guys, they really have it, or this man, this woman, they have it, they will make it. And this man or this woman, no, forget about it. That's not going to work. Was something like that in the air or not?
Bill Eckhardt:Yes. I want to start with an editorial comment. I believe that we should criticize and compare people as little as possible. I do not think that that's psychologically healthy. However, I believe we should criticize and compare ideas. Now you're asking me to mention favorable things about traders, so that's okay, and I'm willing to do that.
I would mention Tom Shanks. He's original and aggressive. Now, in recent years, Tom has been stricken with a grave illness and he's confronted it with heroism. So, I think Tom Shanks is an inspiration.
A second trader I would mention is Paul Rebar, who is intelligent, probing, and he's very level headed. So those two stood out to me.
Rich Dennis admired Curtis Faith, the turtle Curtis Faith, I believe, because he doggedly, Curtis doggedly pursued the trend. So those are perhaps the traders who stood out the most. Although, you know, if you would have asked me about almost any of the traders, any of the turtles, I would have said they were an asset. But I'm just mentioning those people who stood out.
Moritz Seibert:I wasn't expecting any names. It was not my intention to ask for names. It was more like when they were in the room, maybe you would observe that some of them are very nervous or some of them are too relaxed and maybe too nonchalant, and you could therefore kind of say like, well, if you have absolutely no adrenaline or whatever, then it's probably not going to work. That was more the background of my question.
Bill Eckhardt:I see. Yeah, I mean, there were some who dropped out fairly soon, and that indicated that they didn't have the right emotional makeup. There were some who couldn't get along with other traders. I mean, you get a group of people together working in a room, a lot of problems develop. I think we can learn more from the good performers and that's why I mentioned names.
Moritz Seibert:Absolutely.
Bill and Rob, I think we're coming to an end. It's been an absolute pleasure.
Bill Eckhardt:It has been a pleasure too. Thank you.
Rob Sorrentino:Yeah, thanks, Moritz, yeah.
Ending:Thanks for listening to Top Traders Unplugged. If you feel you learned something of value from today's episode, the best way to stay updated is to go on over to iTunes and subscribe to the show so that you'll be sure to get all the new episodes as they're released. We have some amazing guests lined up for you. And to ensure our show continues to grow, please leave us an honest rating and review in iTunes. It only takes a minute and it's the best way to show us you love the podcast. We'll see you next time on Top Traders Unplugged.
