The Gateway to Algorithmic and Automated Trading

Risk Navigator

Published in Automated Trader Magazine Issue 37 Summer 2015

Katy Kaminski considers herself a divergent risk taker. And though this meant that she tried many different things, ultimately, she pursued her love of mathematics on a path to quantitative finance. Now, she researches divergence and convergence, and applies it to financial markets as a director of research at Campbell & Company, a US-based CTA that's been around over 40 years. We talk to Kaminski about theory, strategy and tech.

Katy Kaminski, Director of Research at Campbell & Company

Automated Trader: Why quantitative finance?

Katy Kaminski: I have a background in engineering and mathematics and spent over 10 years at MIT (Massachusetts Institute of Technology). While at MIT, I worked with my PhD advisor, Andrew Lo, a long-time expert on quants and hedge funds. He was actually running a hedge fund as well at the same time. The questions that he asked me to solve were: Why and when do heuristics work in investments? And why do we use simple rules to invest?

It started off with an early interest in heuristics, quantitative rules and systematic approaches. When I was a doctoral student in the early 2000s, I spent a lot of time studying behavioural finance. I really wanted to try to get a good idea of why we use heuristics and rules to make investment decisions.

The topic was considered somewhat too practical for most of my academic colleagues. Now, the topic is very popular, but about 10 or 15 years ago it was considered much more of a practitioner topic.

I started working at the Stockholm School of Economics and after some time there, I wanted some practical experience so I joined RPM, a CTA fund of funds located in Stockholm Sweden.

The fit was really quite natural because being an allocator to CTAs requires that you understand the model that they use, and how to decipher one strategy from another. This requires some understanding of the technical details.

AT: You're also an author. How did you move into writing?

KK: I met numerous different CTAs and started writing articles about investing in CTAs, and understanding how CTAs work. I wrote an article in 2011 that I called "In search of crisis alpha". This became a popular piece which lead to other pieces about CTA investing - momentum investing, systematic investing, futures trading, etc. After some time as an allocator, writer and researcher on CTAs, I went back to academia and decided to write a book on trend following.

While back in academia, I happened to meet Alex Greyserman, chief scientist from managed futures firm ISAM and co-author of "Trend following with managed futures". We both had a common goal to educate investors on why systematic trend following strategies work.

A book is kind of like a quest, we (Alex and I) sat down and said: what do people need to know as opposed to what do we know?

The book was similar to completing a doctorate, you have an ultimate goal and one of the exciting parts is you never know how you are going to get there. We outlined what people should know and divided up the material in certain areas. We knew good answers for some questions, but in others it was less obvious. And it's particularly those areas where the answer was not so obvious where the results became quite interesting.

AT: For example?

KK: This concept of convergent and divergent risk taking. The basic idea is that given what you believe about risk you will apply different risk taking strategies. Often we take our beliefs for granted, so when people step back and start thinking about: "What do I believe?" They may not realise how severely their beliefs about risk affects their choices. Then you can really simply explain why a systematic trend following (or divergent) approach represents one way of taking risk, whereas the convergent approach is another. And then you can easily argue that these approaches work in different risk taking scenarios over time. What has been hard for the systematic space is the black box label.

People think it is some magical box that makes the decision, but systematic trading is rules that we make up that we implement systematically to help ourselves make decisions which are difficult for us to do discretionarily. This is why they are "long divergence" and often tend to do well when making discretionary decisions becomes difficult.

AT: One of the concepts you discuss is the difference between being long divergence rather than long volatility - can you elaborate on that?

KK: If you are convergent you are looking for convergence to a view or convergence between asset classes. You are looking for things to converge in some sort of way. But if you are divergent, you are looking for things to change structurally in the system. This structural change causes the world to diverge resulting in opportunities for divergent risk taking strategies. When things diverge, there is usually volatility but when there is volatility it doesn't necessarily mean the world is diverging.

Most of the theories in economics are focused on the efficient market hypothesis, which is the idea that everything converges and that competition yields to efficiency. If you have that view only, convergent strategies make sense. And some of convergent opportunities will disappear very quickly. The world of efficient markets is a convergent world.

If you go back and challenge the view that the world is always driven to efficiency and think of the financial world more as an ecology or perhaps "market ecology" - similar to Andrew Lo's theories of the Adaptive Markets Hypothesis. From this view, the trading universe and markets are really made up of people, each with their investments and decisions. Most of the time it works very well and competition drives to efficiency. Yet, from time to time we have a structural realignment of this ecology or a stress to the system where everything changes and adapts. And it is particularly when these types of things happen that systematic strategies are specifically looking for those opportunities.

The challenge with divergence is it's very difficult to predict. Let's take 2014 as an example, there was tremendous divergence in markets. In technical terms, there is measured signal in the noise of prices. This meant that heuristic rules could systematically sift the signals out of the noise in prices.

If you believe that markets are efficient, that shouldn't exist. You shouldn't be able to capture that. But if from time to time there are periods where things happen that are outside of what we believe will happen, and things change in a structural way, that take times to adapt to. In these scenarios, systematic approaches like trend following can capture divergence.


AT: Can you give me some specific examples of what you mean?

KK: Just look at oil as one sort of idiosyncratic example. Very few people predicted that there would be that kind of movement in the energy markets last year. And a movement of that type is difficult for a traditional investor to process. If you have a movement like that, where oil is going down and every fundamental model is predicting the opposite, the only type of systems that are going to go against some of these models are a systematic view that is just following the opportunity.

So systematic strategies follow the opportunities based on what the market tells them, whereas a fundamental model requires a view and if the view is challenged it is very difficult for those models to readjust quickly.

Another example from last year, let's think about monetary policy. Divergence is when things are changing structurally and moving in such a way that it may be measurable. The last five years of quantitative easing, what I would call coordinated monetary policy, was synchronising down to zero. And as everything becomes the same that will create very little opportunity for a divergent strategy. That means there will be less opportunities for changes and structural differences across the globe.

Last year, when monetary policy diverged - meaning some areas of the globe were focusing on tightening some were trying to loosen - things were unclear. Monetary policy stopped being so synchronised and what you saw was huge divergence as a result of that.

This sort of unpinning of unified monetary policy caused tremendous whiplash in the FX markets, caused tremendous whiplash in many different markets, and what this does is create an opportunity.

It is very difficult to predict, it is very difficult to model and an opportunistic strategy that just systematically follows opportunities will be the one that may be able to capture that. That is why momentum did so well in 2014.

AT: Is that where we put credit for the upswing in CTAs?

KK: Because the divergence was very high last year. What is divergence driven by? It is driven by structural changes in our financial ecosystem.

High volatility doesn't mean something is happening, it could mean that things are uncertain.

Especially when things are unpredictable. Who was going to guess that oil was going to drop so low? Who was going to guess that monetary policy was going to unravel in 2014 and not 2013, or 2012 or 2011?

These are all things that are really difficult to predict, plus there is opportunity because it is very difficult to react appropriately to capitalise. Being long divergence requires you to follow divergent strategy to capture that systematically.

AT: How are you practically realising these themes at Campbell?

KK: I've known Campbell for a long time, I knew them as an allocator in this space and when the opportunity to come and join as a director in research came about I was thrilled because ultimately I am sitting in a team of quant researchers whose main goal in research is to focus on delivering alpha from diverse sources.

What the convergent-divergent approach helps us with is to better classify what kind of risk we are taking. And what I am doing now is taking some of my past experience in research, especially in style analysis. This is about understanding how we take risks at different levels in our portfolio to understand which risk factors from the CTA trading perspective we are exposed to, and also to help understand to better communicate what we do. Ironically, CTAs are very transparent but investors find them to be confusing.

With Campbell's approach, a main objective is diversification for our portfolios. This makes sense when you are a divergent risk taker because what we want to do is instead of trying to have one strategy, we strive to have as many strategies that make sense and that have an appropriate Sharpe Ratio contribution.

We do things as a process: we start with models, we have to put the models together, we have to aggregate our trades and (then apply) what we call investment discipline.

I always attune what we do a little bit to building a cell phone. You build an interface and the interface has to be robust. It has to take shocks appropriately. It has to adjust to different environments, different inputs, different scenarios. And the output is our trading results.

The investment objective or mandate for our flagship investment approach is enhanced trend following. What I mean by that is that we have an 80% risk allocation to trend following, and trend following is what? It is divergence. We have a large exposure to divergence and we try and capture divergence in as many ways as we possibly can.

On the other hand, we also have an exposure to non-trend, which is actually convergent. In the convergent bucket, we are looking at things like mean reversion, carry trading, macro effects in markets.

We have somewhat of a barbell portfolio in that we have a large exposure to divergence but we balance that divergence with some convergent trading in futures. 20% of our risk allocation is convergent, and this concept has been I think one of the reasons that Campbell has performed very well over the past few difficult years for CTAs pre-2014.

One of the main things I try to tell people in my presentations is: you need to balance both convergence and divergence. A lot of people come to the futures space for divergence and our portfolio balances both.

In periods where divergence really suffered, we had a really broad range of models that worked in varying scenarios. It is kind of cool to see that the portfolio at Campbell is directly in line with the whole concept of convergent and divergent strategies - that you should blend the two to have a better overall risk adjusted return.

AT: Why 80/20? What went into that decision?

KK: Futures are highly liquid, highly competitive. Convergent trading in futures has more capacity constraints, momentum is and has always been the primary return driver for risk premia in futures trading. So given that we have traded futures and we focus on opportunities in futures markets, it seems very natural to focus on what is the great advantage of trading futures? The liquidity, the opportunistic ability to trade. The easy access to leverage all those great characteristics of futures is really the best way to capture momentum. So we do actually believe that non-trend is a very good component.

Now, this is our main core flagship objective but ultimately we are not confined to that. We have investors who want X, Y, and Z and that depends on what your objective is as an investor. We find that most people come to the futures world looking for the momentum premium and we think that the 80/20 is a very logical balance between the two because there is another point - we net models between trend and non-trend.

Remember that trading trend is a much more absolute directional strategy. A lot of non-trend strategies can be more short term, and they can also be somewhat relative and more idiosyncratic. The balance in terms of risk works very well with a smaller allocation to non-trend.

AT: Can you tell me a bit more about Campbell, and what kind of technology underpins the trading strategies?

KK: One of the exciting things about Campbell is that we advise the longest continuously operated commodity pool in the United States. The firm was started in 1972, so it has been around for over 40 years, longer than me, which tells you something about how old I am, but longer than I have been alive.

Campbell is focused on futures and has been historically very successful in quant strategies. Right now we have upwards of $5.4 billion in assets under management. We have investors in both the private wealth and the institutional space. We have over 120 employees, in research we have 25+ researchers, and 30 to 40 people who are primarily in support of the research faculty - so data engineering, systems development. It is pretty awesome because when you are an academic you have to do a lot of that stuff yourself. Here we focus on idea generation, implementation and research. If I want to test things, the infrastructure is so sophisticated so I can just ask for a certain data series and I get it, clean.

One of the major advantages of being in a firm like Campbell, it has been around for so long, the understanding of technology is incredible. Ultimately the majority of trades are executed via algorithms, but there are nine dedicated traders covering 24 hours of shifts across Asia, Europe and the US.

In terms of trading infrastructure it is really a world class environment. You have the research teams, which are more idea generation. Then we have engineering and data teams, down to trading implementation and execution. It is almost like a streamline from the research, where we nerds like myself are sitting around asking what is the next idea, and then we're down all the way into the implementation and production quality of the code.

There is a unified framework which aggregates all of our positions, and we have an execution system which takes into account all the trading costs. It's very sophisticated and I think that is exciting to be part of a whole process like that where you start from an idea and it goes all the way down to an executed automated trading system.

AT: What are some of the signals you are looking for to know your strategies are on the right track?

KK: We have seen divergence in markets and it's very exciting to see what is going to happen in markets going forward. Right now, my current research is on style analysis and portfolio risk management factors which differentiate models from each other - things like exposure to liquidity. My goal is to both provide new perspective internally so that we can continue to develop how we execute risk management, find ways to improve our processes and also externally improve the way that we can compartmentalise and explain what we do to investors.

AT: What do you think is the most overlooked issue in investing today?

KK: There is a lot of discussion about the importance of new markets. The role of futurisation is definitely an interesting question. What is the evolution of futures markets and futures trading as volumes in futures increase?

AT: When you say futurisation, you mean swaps?

KK: Yes. In the United States some of that has been implemented, but in Europe, EMIR is still chugging along. But I think the long term impacts of this is going to be interesting. I think it will have an impact on how markets trade.

There has definitely been an augmentation of volume in the futures-like structure and there is more of a blurred line between the way that OTC contracts were cleared and executed in the past. This blurring of lines is pushing more and more towards the futures structure, and it is interesting to think about how that is going to change the way futures trade. Which contracts are going to become liquid and accessible? How is HFT going to manoeuvre into that space more aggressively? Or not? How that might affect us?

Futurisation as a concept is something that is happening behind the scenes and the playing field is changing slowly. We may not see that until we look back five years and think wow, I can't believe that this happened.

So that is one important question, looking at new markets, how are the dynamics of markets going to change with this structural change in markets?

I also get a lot of questions about when divergence is going to be over. Again, divergence is not predictable so if you believe that the world is going to stay in some turbulent uncertain state dealing with low interest rates, among other things, then I don't see any reason why there is not going to be opportunities for trading going forward. If the world goes back to unified monetary policy and very synchronised views, then I think it will be difficult again. But I don't personally think that will be the case.

Divergence is not something that you can value. It is not like stock where you can say, well here is what we think the value is. It is really sort of an opportunistic way of taking risk and you have to have opportunities for that to work. If there are opportunities it will work, but if the world becomes very boring then that will not be the case. There is no way to control that phenomenon.

AT: One last thing, have you ever faced any obstacles along your career path? I ask because I have heard some real horror stories from women in the trading world.

KK: I have my fair share of horror stories. I have always done what I loved and just tried not to care what other people think. Sometimes I remember and it does affect me, but I've loved math since I was 12 and I just never let anybody tell me that I couldn't do something.

I don't know if you have read the book "David and Goliath" (David and Goliath: Underdogs, misfits, and the art of battling giants by Malcolm Gladwell) - your greatest weakness can be your greatest weakness or your greatest strength. I see being different and having to fight hard for what I do as a great strength because when you have to adapt and learn, you are stronger as a person.

My first experience working in CTAs was in Swedish (a foreign language) and that was interesting. The students voted me teacher of the year, and I was possibly one of the only women, and the only foreigner, that has ever won that title. If you asked me five years before would that ever happen, I would have said: fat chance in hell.

I think that is the point, there have definitely been obstacles, but I always focus on doing what I like and what I love. It is more like value investing, focus on the value, the grooves, not on the distracting things.