Our next show provides you with the opportunity to learn from a highly educated founder and fund manager.

He Studied Economics at the Luzon Universidad de Carlos III de Madrid. He went on to earn a PhD in Quantitative Finance in Evolutionary Finance at University of Zurich. Upon graduating he agreed to a research position with Zurich Capital Bank.

Horizon21 made an offer to have Mathias and his business partner Dr. Tilman Keese build a systematic trading program. In 2010 they left Horizon21 to go out as entrepreneurs with AllMountainCapital.

Please give a warm welcome to, Dr. Mathias Bucher.

In This Episode, You’ll Learn:

• The story of founding AllMountainCapital and how much AUM they currently manage
• How they outsource all non-core aspects of the business so they can focus on Research, Trading & Client services
• On the changes in the CTA industry from 2007 to the presentWhy central bank actions are correlated with a drop in volatility since 2009
• The nature of the AllMountain trading model and how it has coped during challenging times
• About the Modules that make up the AllMountain trading program
• Sectors and markets that AllMountain trade
• How their different system works and why they use it the way they do
• How they quantify trend strength in a market

Resources & Links Mentioned in this Episode:

• Larry Hite and ISAM
• Watch another interview with Dr. Mathias Bucher on Opalesque TV

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 “The Many Flavors of Trend Following” 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 so more people can discover the podcast.

Niels

Niels

Mathias, thank you so much for being with us today, I really appreciate it.

Mathias

Well Niels, let me thank you first for having this conversation with me tonight. I think it's actually a really great idea of yours to do this pod cast series. It should, I think, certainly help investors in gathering relevant and easily accessible information, if they want to invest into CTAs. So yeah, two thumbs up for that.

Niels

Excellent. Thank you so much. Mathias, I think it's fair to say that you belong to the newer generation of CTA firms, who often come at the trading world with a deep academic background, and not like some of the old legends in our industry who often started out as a discretionary trader, kind of a by-the-seat-of-their-pants trader, so perhaps a good starting point would before you to take us all the way back to the beginning, telling us the story about how you went from the stress free world of academia to the high temper world of global finance, as well as the transition from being an employee, early on in your career to now being an entrepreneur, what inspired you to make all these leaps and how has All Mountain evolved since its inception?

Mathias

Right, your assessment is certainly correct. So as a short introduction to myself, I started into economics at HEC de Lausanne, and at the Universidad Carlos III de Madrid, and after having finished my studies, I first went to strategy consulting working for McKinsey & Company for a couple of years. During this time my wife was in the high yield business and I gradually discovered that finance is actually, in many ways, more interesting than what they teach you in the college classroom. Finance actually brought me to the point where I decided to do a PhD in Quantitative Finance, and more specifically Evolutionary Finance, and I did so at the University of Zürich, with professor Hentz. During this time I also worked as a researcher to Zürich Kantonal Bank.

Niels

Exactly. How long did it take for you and Tilman to build the program, and how did you come up with the initial ideas, because you were obviously coming straight from academia and, I'm guessing, at this time probably didn't have much real trading experience?

Mathias

Niels

And at that time did you look at any other CTAs trading in our industry? Did you look at any of them to kind of have a sense for what they were doing and where you wanted to be different, or was it really just based on the research and experiences that you had from your own?

Mathias

No, no, absolutely, we clearly were aware of the traditional ways that CTAs were built: time series, moving averages, volatility breakouts, all that. We were also conscious, at this time that we needed to have an edge, we needed to differentiate ourselves. That was clear from the beginning, and we had quite a good intuition of how we wanted to be different. So that was, essentially, our starting ground.

Niels

Interesting. Perhaps you could give me a brief overview of the program you run today: when it started, and how much AUM you have in the program?

Mathias

Niels

Excellent. Fantastic. Well before we jump into the detail of the program, I just wanted to jump to a slightly different part, and that's a little bit about your company and how you've organized it. Clearly today there is a lot of technology available, there are a lot of firms today that offer services that allow smaller firms to outsource. How have you gone about dividing between what to keep internal and what to outsource, how have you done that?

Mathias

Right. You know, Niels, coming from this strategy consulting background, for me this was, of course, an exciting challenge, to build our own company. We had, from the start, a very clear view what segment in the value chain we wanted to occupy - where we wanted to focus on, and this is clearly research, it's trading, and it's client activities. All the rest, we designed our firm to outsource to high quality specialized partners: be it in the whole product management, be it in the IT infrastructure part. This was not up and running from the start, but today we have service on three continents, fully encrypted, etc. etc. all the bells and whistles. This, of course, you cannot achieve in a new company, with limited resources, you need to rely on a strong partnership with external parties.

Niels

Yeah, I agree, and I think almost in the sense it's like smaller newer companies have a slightly competitive edge, because I do see that some of the larger, very well established firms that have been around for a long time, do carry some quite heavy infrastructure simply because their systems were built at a different time when technology was not as available as it is today. So I agree with you, you can get a lot done with smart thinking and adequate strategy. In terms of the program itself, do you have any target or optimal size that you think you're striving towards?

Mathias

We don't have real limitations in terms of how much we can trade. We consciously trade only the most liquid futures globally, so this gives us no limitation in the foreseeable future.

Niels

Mathias

Niels

Well, absolutely. I think you are referring to the study from Larry Hite, at ISAM, which is obviously very interesting, and great to see that these firms are able to go back all these years, these centuries, really to get data, and in fact that they data confirms what I think at least practitioners of the CTAs strategies believe, namely that trends will always be there, and but it doesn't mean that you make money every year. So yeah, very interesting indeed. Thank you so much for sharing that.

Now, when you look at your own track record, in order to set the stage before we jump in and talk about the program itself, is there any different periods, different stages of the track record, meaning, have there been any major discoveries or upgrades to the system that people should be aware of when they look at the track record? What I mean by that is if you look at someone who has been around for 20 years, you can be sure that the models they trade today are not the models they traded when they first started out. So I'm just trying to set the stage here, to find out whether you have evolved the model a lot, before we talk about the model itself.

Mathias

I think it's a super relevant question. The nature of our model has not changed. This said we can talk later about the reasons for that. I think we believe a lot in robustness, however, there have been a lot of different steps and measures have been taken, a lot of research effort has gone into many aspects of how we trade, for example, the portfolio composition has clearly become better. The risk management has improved. The efficiency of execution has been improved. We have almost fully automated all processes. These kinds of things have dramatically improved, of course, but we strongly believe if we commit ourselves to be a medium term trend follower, we should stay a medium term trend follower, and not start trading intra-day patterns or this kind of deviations, and allow for style drift in the end.

Niels

Sure, makes sense. Now, for the trading program itself, maybe you could explain in your own words how you have stuck to the program and why you've designed it in the way you have. Is there a particular philosophy behind it, etc. etc.? Just in your own words how you best describe what you do.

Mathias

Sure. Our trading program has essentially two modules. The backbone module is a medium term trend following module, and there is a second module that kicks in that is situationally dependent - which is a mean reversion module, that essentially protects the invested capital if you see strong volatility increase against the trend. So these are our two modules, core modules of our program. Now let me talk a bit about the trend following module.

Niels

Sure.

Mathias

Maybe I could start talking about this by looking at our typical trend following model would look like. It will consider time series information: looking at moving average crossovers, volatility break outs, and create a portfolio from the bottom up. So it's time series information and bottom up creation of the portfolio according to when these trading signals happen. So we take a different approach here. Based on our quantitative background, we essentially try to rank the markets in our universe, and then pick the qualitatively best trends, and selectively combine these best trends into a portfolio that will not encompass all markets of course. It will typically have 10 to 20 positions so we combine them into a portfolio by achieving a second goal, or trying to achieve a second goal, which is optimal diversification of the portfolio positions picked. So we strive to have a selective portfolio of not very many positions - essentially picking the best trends out there, and combining them into the portfolio in a way that we have an optimum de-correlation of these candidates.

Niels

So let's break that down a bit, because that's a big mouthful for people to comprehend. So let me just start by setting the scene a little bit. Perhaps you could mention, not individually, but perhaps you could tell us a little bit about the markets you do trade, how many, and also the sectors, whether you are fully diversified, let's start in that area.

Mathias

Sure, sure. As I mentioned before, we trade all future markets globally that are liquid enough to trade. So we never want to run into any liquidity problems, so we constantly monitor the liquidity of these markets, and so, at the moment, we are looking at about 80 to 90 markets globally. We trade all sectors: equities, bonds, meats, metals, Forex, softs, grains, we trade all sectors, and we typically will strive to have picks from all sectors at the time. This will be a major contributor to this diversification that I mentioned.

Niels

Sure, there's no doubt that sector weights plays an enormously important role in determining performance, at the end of the day, so that's a very important point. Let's go back to the model itself. You mentioned that the core is a trend following model.

Mathias

Exactly, a medium term trend follower, but instead of using a moving average that will give me a signal when to enter, I will apply one econometric model to create a score for each market in the whole universe at all times. So we are looking at markets cross-sectional

and we use one model because in the end, for robustness reasons, to come up with this core and this core will tell us, (as a hypothetical example) in this market copper has a great trend right now up, bonds are strongly up, and equities are up, so we will have the same score for all markets cross-sectional, and this allows us to compare these markets. This is our first step that we calculate on a daily basis. We compare all markets on a daily basis.

Niels

But what I'm actually interested in, Mathias, is just to take a step back because, I think most people listening to us today are used to thinking of trend following as being, as you rightly say, it's a crossover of two moving averages. It might be a price breakout of a channel where the market, if it goes above the last 50 day high, it's a breakout. You talk about the signal generation in a different way, and I'm not sure I actually understand what you mean by a model like that. So if you can explain that a little bit more as to how does it know that copper is in a big trend if it doesn't look at the price breakout or the moving average crossover?

Mathias

Right. Let me maybe take one step back and let me explain the motivation of why we use this different approach. If you use time series signals, moving averages, volatility breakouts, you will not be able to control the amount of markets that you trade - the signal happens when it happens and you trade it. So your portfolio typically will be rather large and will have many positions, maybe 50 positions at the time. We try to be more selective. We want to be invested only in the 10 to 20 best markets at the time. So we need a methodology for how we can compare the trend quality at a given time, cross-sectional, across all markets. So what we have developed is an econometric model that essentially scores the trend with various criteria that I won't go into right now. It scores the quality of this trend for each market.

Niels

Just to clarify here. What you need in order to make that score, is that based on anything other than the price of the market?

Mathias

No, it's absolutely only price driven. That's totally right. So we have a single score for each market, and some scores will be high, some scores will be close to zero and some scores will be very low. So the positions that have a very high score will be picked long, because there is quite a big probability. The positions that have a very low score, will be picked because there is quite a high probability, and the scores that are close to zero are likely not to be picked.

Niels

Sure, and the inputs that you need, i.e. the price, is that just a daily sample. So at the end of the day you need to calculate this, or is it something that happens during the day as well.

Mathias

No, that's right we use high, low, open, close prices to calculate this score. I think your calculation frequency should also be in relation with the span that you are looking at. We want to have a program that has characteristics of a medium trend follower, so it does not make sense to recalculate these score more than once per day. You can do so, but the difference in the score values will be so minimum that it essentially has no relevant impact.

Niels

Sure, sure. So you get this score and how does the model know whether to pick the top 10 or the top 20, is there something that dictates, saying OK, if it's above a score of 70 or below a score of 30, just to pick two numbers, we take all of those signals and we leave everything in-between alone, or do you set other parameters that decides the outcome of how many positions that you end up with?

Mathias

Niels

Sure, OK, excellent, and does the model know, at this stage, when it brings you on a daily basis, say the top 20 markets that you should be involved in...how does it go if there is a new market coming along tomorrow that actually gets a score that is high enough to warrant a position...how does it know how much to risk in that trade?

Mathias

Yes, so essentially what we are looking at here are thresholds that need to be breached, so when you have just a minor change in the relative ranking, it would not justify to replace this position because of trading costs, slippage, etc. etc. So we want to see a distinctive outperformance of a relative score to change one market against another.

Niels

OK. When you get in a position, does it automatically mean that you have a stop loss associated with that position, or do you need the model to basically kick a market out of its preferred list status in order to get out of that market?

Mathias

No, essentially what you are talking about here is our risk management approach.

Niels

Well, more the exit side. I know it ties into the overall risk management which we'll probably talk about also on top of this, I was just trying to get a feel for whether you treat your positions individually, meaning that I'm going to allocate maybe 50 basis points risk, so if I lose more than 50 basis points on this position, I'm out, or whether there is a different mechanism that actually decides how much risk to have.

Mathias

So there are two ways that our program can lose a position. Either it's through risk management, and I would count the stop losses as a part of risk management; or it can get crowded out, so it can be replace with a better score - let's say that it lost the space in the portfolio for this asset class, so it will get crowded out. So there are two ways of losing a position.

Niels

What I'm sensing from our conversation is that the model is really somewhat different from a traditional trend following model, because it's not really something that would go from being long to being short in a market. It's really just looking at the overall universe of markets, deciding which of these opportunities should be in the portfolio at any given time.

Mathias

Correct, yes.

Niels

Excellent. If we just look from an overall point of view, what does that equate to when it comes to say, trade duration? How long are you typically involved in a winning trade, or in a losing trade, how does that all fit together?

Mathias

Niels

Mathias

Niels

And how does the system know to turn up the risk budget, meaning just for clarification, if you want to add a new position, and it has a certain score, is it the score itself that allocates...so just as an example, say it gets a score of 65, does all positions with a score of 65 get the same risk allocation, and maybe a market coming with a score of 75, it gets a slightly higher risk allocation, or how do you decide how much to risk in any new opportunity?

Mathias

So there are two criteria: there is the pure score, which tells us this is a good opportunity; and then there is the diversification function in the existing portfolio. So two criteria here. It's also very important to mention that once we have picked our candidates, we will not size the positions according to the score. Score has to function to identify opportunities, but the sizing is driven by risk management. It's not driven by scores. Scores is more the flag that waves, "Hey!" This is an interesting market to get involved in. The sizing will be done by risk management.

Niels

OK, interesting. Now, we are going to talk a little bit more about that for sure, but in terms of performance drivers, if I can call it that, what do you think is the key performing drivers in your model? Clearly selectivity seems to be very important, but is there anything else that you would say is part of your difference that you can point at?

Mathias

The performance driver is the volatility expansion in the market, and our model is able to identify volatility expansion situations well, and then trade it selectively. And then we have, of course, a pretty good risk management on top to mitigate the risks that are associated with it. But it's really very important to state and understand that in the market, (and in a bull market) where volatility is collapsing, you can have as good a risk management as possible, but you will not be able to make a system like ours profitable. You would need to trade markets differently.

Niels

And so once you are, for clarification, when you are in a position, so do you change the position size during the lifetime of the position, or does that stay constant until it gets kicked out of the portfolio.

Mathias

Again this is a function of risk management. If the overall risk increases the position (if nothing else changes) would increase too. If risk management perceives danger for this position it gets, maybe, totally cut, or it gets reduced. So the sizing is always a function of risk management in our model.

Niels

OK, very interesting. Trade implementation is a subject that I wanted to also touch upon, and that is a little bit about how the system actually runs. How difficult or easy is it to run and maintain? Because I think, for a lot of people, this may sound like a black box, and so on and so forth, but the reality is, of course, that people using these types of systems they know what goes in and they know what to expect to come out. But just sort of the operational side of running a model that follows so many different markets and has a few different parameters, how does that work in practice?

Mathias

So we, from the start, went to great lengths to fully automate and only trade electronically. It's very important in our set up, in our case, so our signal generation is fully automated. Our stops are fully automated, and we trade everything electronically, algorithm based, so we did, essentially...a key ingredient to running a model like ours successfully over time. It mitigates a lot of risks.

Niels

So does that mean that nobody needs to watch the computer, or what does actually happen?

Mathias

(laugh) No, of course not. It's totally an illusion if you think you can program a system and then let it run. That would be fully not responsible behavior. You could never do that in a real trading environment. Of course, monitoring is up and running. It means that we don't need to intervene unless there is a hiccup somewhere. It can be that somewhere in the internet an outage occurs or that we want to trade a market and there is a market outage or market feeds are not coming properly in. You know, API feeds for example, these kinds of things need to be constantly monitored and, again here you can rely on a lot of technology, you can get alerts via email, pop-ups, etc. etc., and we fully use this opportunity.

Niels

But I guess what you're saying is that actually the data will come in automatically, the system will make its calculations automatically and it will send any adjustment orders automatically. Nobody needs to press any buttons on a daily basis.

Mathias

This automatic trade entry happens in case of stops, just because it doesn't make sense that you wouldn't need to review this. However, in the case of portfolio adjustments, we always have this "four eye principle", that we manually oversee, before the order gets executed, and we verified that it's in line with the whole program.

Niels

OK, so you get told every day whether there is an adjustment amount?

Mathias

Correct.

Niels

Now you mentioned another subject which I would like to talk about next - risk management. You've mentioned it and talked about it so that my understanding, at least, is that this is very important in the way you approach things, the way you design things, so maybe you could tell me how you define risk and what targets of risk you're looking at, and how you've gone about using this approach?

Mathias

So, we're typically looking at risk as the volatility with the VAR budget, like pretty much in line with the industry here. Our risk management consists of three layers...

Ending

Thanks for listening to Top Traders Unplugged. If you feel that you have 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 that our show continues to grow, please leave us an honest rating and review on iTunes. It only takes a minute and it's the best way to show us you love the pod cast. We'll see you next time on Top Traders Unplugged.