The Gateway to Algorithmic and Automated Trading

Riding the momentum. Trials and joys of setting up a systematic trading shop

Published in Automated Trader Magazine Issue 34 Q3 2014

Since starting in the securities industry almost 20 years ago, Lars Wind has seen a number of crises come and go - the Asian crisis, dot.com bust and the global financial crisis notably. He's worked all over the world developing bespoke strategies, ultimately coming to London to start up his own fund QLO Capital.

Lars Wind

Lars Wind

Chief Investment Officer

Automated Trader talks to Lars Wind and his team about the bumps along the way, details about the strategies and why he thinks CTAs are set for an upswing.

Lars Wind: I started out in the securities industry in 1996, one year before the Asian crisis broke out and that was my first interesting macro event. It started shaping my thinking about how we should approach markets because it was a major dislocation. I moved on, covering the TMT (telecom, media, technology) sector in 2000, which was another very interesting, not macro event as such, but an event where you have the crowd herd behaviour dominating markets more than fundamental behaviour. In 2002, I moved to the buy side for ATP, the Danish government pension fund and joined the equity strategy team. I built ATP's equity derivatives business in 2002, there was a lot of hedging activity going on in the life and pension section and then I gradually progressed through the system, doing equities strategies and TAA (tactical asset allocation) and was promoted to head of TAA in 2004. Back then, it was a very traditional pension fund approach. We would go into a boardroom and have a debate about our views of the world, it was a very inefficient way of making decisions and there was no way of tracking the real value added of the various people around the table.

Automated Trader: Why do you think this was inefficient?

Lars Wind: It was the lack of transparency. Essentially, people from the equity division inside ATP, the fixed income division, and we would argue to try and reach some form of consensus decision involving the CIO, even people in the back and middle office would join in. We really had no transparency of what the overall strategy was. We weren't really making any very good decisions, so I suggested that we completely restructure in an overlay approach using the derivatives that we now had. I got a mandate to launch the TAA macro fund internally, which I did in 2005 with a team of four people. Instead of people meeting from different divisions it was a dedicated team responsible for implementation.

"…the correlation has dropped tremendously and that is helping…..in effect the market has gone back to where it was prior to 2006."

Automated Trader: What did your trading strategies look like then?

Lars Wind: It was completely discretionary; we traded for about a year and during that year came into contact with ADIA (Abu Dhabi Investment Authority). I met Jean Paul Villain, now ADIA's head of strategy and he was very interested to know more about this overlay approach to macro investing. I moved to Abu Dhabi in 2006 and had a dual mandate. On the one hand I was responsible for setting up an internal hedge fund desk with ADIA, within the tactical asset allocation unit, not the alternatives department, which is typically the CTA business - that was one part of my job. The second part was to act as an adviser towards the investment committee inside ADIA about medium and long term asset allocation. That was very much traditional, the same approach we had seen in ATP. But where we did have complete discretion was with regard to the internal hedge fund desk and that was exactly the philosophy they (ADIA) wanted to create - transparency in terms of what the strategists were doing. My job was to set up a desk where each strategist would run his own internal fund with complete discretion to implement his own views. It was during that time, in the first few months of 2006, that I really started studying CTAs in great detail. ADIA is a very large investor in the CTA space and they have been since the 1980s. Conceptually, it made a lot of sense to use systematic trading to express macro views because we know that the markets get dislocated at times.

Simon Hampshire, Business Development

Simon Hampshire,
Business Development

Automated Trader: How did you then put systematic strategies in place?

Lars Wind: We saw dislocations in '97, '98, 2000, and we saw it again in 2006, the commodity boom, in '08 the credit crisis. And I realized that it is actually possible to build a vastly more diversified system just using price signals than you ever can as a discretionary manager, just because of the economies of scale. So I decided that the systems would form a core part of my systems trading philosophy, but I also always felt that it was important to have a discretionary overlay which is very much part of what we do (at QLO Capital) as well. While we believe that it is possible to build good systems, the philosophy that we have is that first of all we only trade trends that make sense in the context of what is going on in the real world. We are not trying to look at tick data or anything like that and using this discretionary framework we can certainly add additional value by constantly monitoring the exposure we have in the quant book, and then, using a fundamental approach we can critically go in and look at the positions from time to time and adjust some of the risk exposure we have in the quant book.

Automated Trader: Can you be a bit more specific about how that works?

Lars Wind: The philosophy on the quant side is price driven, there is no fundamental input. Whereas on the discretionary overlay it is very much fundamentally driven. I am an economist by training, so we use models that are fundamentally driven but we don't apply them in a completely systematic fashion, they shape our thinking more than anything else. Whereas on the quant side, we always go with the signals, we always implement everything in the quant book. What is unique about our approach is that we designed the discretionary overlay so it has a negative correlation to the quant book, it acts as a diversifier by design, meaning that we will never amplify the risk that we have in the quant book. It is sort of roughly split 80% systematic and 20% discretionary.

Automated Trader: How does that play out on how you trade financial products?

Lars Wind: Rates, particularly in the US and UK, are going to move up, I think that is not too controversial. It hasn't happened this year, but the way we approach that is; given that we are predominantly trend following on the quant side, the way we then go about and approach the markets is whenever we see bond futures, particularly in the US and UK, rally we will go in and build a short position in the discretionary book to partially offset the exposure we will have in the quant book.

Betina Wolf-Andersen, Head of Trading

Betina Wolf-Andersen,
Head of Trading

Automated Trader: How did you get from ADIA to running your own fund?

Lars Wind: After two years with ADIA I came into contact with Omni Partners, which is a special event fund here in London. The CIO Steve Clark agreed to sponsor the strategy and set up an independent structure. I moved to London in January 2008, and that is when Betina (Wolf-Andersen, QLO Capital's head of trading) joined from Centaurus Capital. We started trading initially in London but then relocated to Geneva in April. We traded out of Geneva for about a year, it was a great year for CTAs performance-wise but it was a very tough year to raise capital. All the smaller funds were shutting their doors. My former employer, ATP couldn't make external allocations back then; their hedge fund business was still very young at the time. They offered us a substantial internal ticket, initially $200 million, a big step up from the $20 million we were trading in Geneva. So we joined ATP in Copenhagen in 2009, taking the book up from $20 to $200 million in a single step and then gradually got increased allocations up to $450 million, which is the internal maximum, over the next couple of years. Then, [when] ATP decided to shut their hedge fund business at the end of 2012, Betina and I made a decision that we were going to try and set up our own shop again.

Automated Trader: Why did ATP shut down the hedge fund business if it was doing well?

Betina Wolf-Andersen: I would describe it as a political decision. The way hedge funds are perceived in Scandinavian culture was not sustainable with a state pension fund. We don't even have a proper word for hedge fund in Danish, that gives you some indication of how it is looked upon and not really properly understood either. The product itself was very profitable and successful and had a huge amount of investment, managed to stand on its own two feet successfully for quite a while; so it was a shame.

"You can have a great model that works, but it has to work in the context of the real world."

To go back a little bit, when we were in Geneva it was a difficult year to raise capital, but it was also a difficult year for the world and we made money, but it was hard work and a lot of focus on setting up the fund. The whole world was upside down for most of it…in the environment, there wasn't interest in investing in anything and basically there was liquidation left, right and centre. So, going to ATP and getting a big ticket, and not having the investor relations aspect of running a fund, and getting the chance to scale our fund up and run it in a big size, (proved) it is scalable. Because that is one of the question marks people have over small strategies, are you scalable? Well yes in theory, but we would, in practice, show people it is fully scalable from $20 to $200 million. It makes no difference to us in the way we think or run.

Lars Wind: So we moved to London in April 2013 with a view to set up our own shop completely independent. We did have a sponsor, but it was a small company so we had a couple of months going down the wrong way, when we realised that the institutional requirements today are just a lot higher now than they were back in 2008. We were never going to be able to offer the institutional framework that is required to get the bigger tickets and started looking out for more of a platform and Simon (Hampshire, principal at QLO Capital) joined us.

Automated Trader: And Simon helped with finding Trium as a platform?

Simon Hampshire: The idea was to get a business going. I had been running the business at HSBC for futures and Lars and the team had been one of the fund managers I was responsible for looking after. I had done my due diligence on them for four years internally, knew how they traded - very impressive pair of PMs (portfolio managers). They had the credibility, the track record, everything worked well from my point of view to give me the reasons to go in with them and join the business. The issue then was: how do we go about setting up the business? The limited amount of money needed to be used for trading, and you dilute that by bringing up a whole load of additional people that you have to salary for - compliance, regulatory and risk management office space and back office systems. One of my ex-colleagues is one of the partners here at Trium Capital and we decided this was a good way for us to have everything set up for us that we needed to do our business without having to pay huge amount of costs. We share the costs for the infrastructure here.

Automated Trader: So Trium Capital operates like an umbrella for funds?

Lars Wind: It is an alliance and more than your standard platform because Trium will only onboard uncorrelated strategies.

Betina Wolf-Andersen: Investors question about Chinese walls, the whole internal sharking and where you get very secretive - which is not a productive environment. They have people that can share infrastructure but not have the competitive field where you have to be protective as well. We made the deal with Trium in October 2013.

Automated Trader: As far as I know you just got your first commitment?

Lars Wind: (Getting the licences and approvals) allowed us to get a little bit away from London, which is very tough at the moment, and into the US, which is actually looking better. There is also a longer tradition for CTA investing out of the US. We received a commitment from Jeffries (in June); that was the first ticket and hopefully we should be on DB Select in a couple of months.

Simon Hampshire: By the end of the year, we expect to be at $50 million; that is our target. So, starting with the $1 to $5 million tickets and building up to the $10 to $20 million. It is very difficult to leap unless you have an initial commitment where somebody buys you out in a group.

Betina Wolf-Andersen: But it is also institutional requirements that they are not allowed to be more than 20% or 25% of any fund, so [before] they give you a big ticket, you have to be quite big already. It is a bit [of a] chicken and egg scenario in that respect.

Automated Trader: How are you working around the difficult environment for CTAs?

Lars Wind: It has been difficult going in the commodities space, but we have seen a major turnaround this year and we are up 20% year-to-date. About half our models are what we call conditional models, or selective trend following and the other half are what we call pure momentum models. What that means is we have a core trend following framework in a very traditional sense - you play all markets at all time frames, at all times - and we have another group of models where we selectively try to identify which product we should trade and that is working incredibly well right now.

"One of the most important things is to check for consistency in data, and that data that you get daily is actually correct."

Essentially what has happened is the correlation has dropped tremendously and that is helping, and that is a very sustainable thing as well because when you look at the asset class in '06 to early 2008, you had a massive move in the entire asset class and it was fairly easy to make money - you just had to be invested anywhere in this space. Then you had a huge collapse, it was very easy to make money if you played the short side. But then it became very difficult because of super-high correlations and absolutely no trends (between 2011 and 2013). What has happened now is that individual stories are certainly coming into play again, so in effect the market has gone back to where it was prior to 2006.

Betina Wolf-Andersen: For me as a trader, it also means that there is value to add on the discretionary side because you can focus on the themes and the little stories, and you can use those to shape the way you think of trading in terms of the quant book. The other thing that differentiates us is a slightly discretionary approach to the way we execute our quantitative trades. For a lot of the other pure CTAs, this is an automated process - generate a signal back into the machine, shoot it off, gets filled, book it in and off you go. We run all of the models to get signals and we do trade off these signals, but in terms of when and how there is a human input - as in myself or Lars, or both of us - determining what the ideal moment (is) to be in this particular trade. The fact that you can focus on the individual stories and know the things underlying, helps you shape the way you think about the trade and the price action.

Lars Wind: We still put it in a control framework, so the way it works on the execution side is that Betina has a 24-hour window from whenever we get a signal until that trade has to be put on, so there is discretion during that 24-hour window to put on the trade and we obviously measure that, we run weekly reports where we measure the value add of that process.

Automated Trader: What frequency do you trade at? What are your holding periods?

Lars Wind: Our average holding period is about 20 days and we have about 1,500 trades a year, but only about 500 real trades, which is round-turn trades - so the rest are rolls and variations in the strengths of the signals. So, relatively low frequency and again, it ties into the fact that we invest in the trends that make sense in the real world; it is not super high frequency, we follow the story from a fundamental angle as well.

Betina Wolf-Andersen: When you sit with the models and you see what positions you have in there, you find stories that are maybe not reflected and the discretionary part is nice that you have the freedom if you feel the models have not captured those. Or other relative plays that might be interesting, that the models, being straightforward trend following, have not captured either.

Lars Wind: In other words, there are two types of trades in the discretionary book and that is what we call hedging trades and uncorrelated trades. Hedging trades are exclusively focused on reducing risk from the quant book when it makes sense from a fundamental point of view and the second group, the uncorrelated trades are trades that are not reflected in the quant book and do not amplify our risk. Then, there is the additional element of discretion that exists within the 24-hour window within the quant book, which is more a question of timing the trades than anything else.

Automated Trader: What kind of technology underpins QLO's trading strategies?

Lars Wind: We are fairly low tech. It is all based on running Excel models and we want to open them up and look at them every single day. It ties into a very fundamental approach so we can evaluate it on a daily basis. In that regard, it is not particularly sophisticated.

Betina Wolf-Andersen: We still haven't found that there is a realistic alternative to Bloomberg data, and the data feeds and also our models and the way they are built with Bloomberg codes, the idea of changing to another data provider still untested is a very daunting prospect.

Lars made a lot of the models, and I joined him from a different background. As an outsider coming in, and looking at Excel models, some of them are quite substantial and complicated but because they are in Excel it is possible to work your way through, and you see where all the different feeds come from and how they actually work. Even if you are not particularly technology-focused, if you take time to understand how it works - so waiting for this price signal, these are the thresholds, these are the products, from that point of view Excel models are quite good because they are very hands on. When something happens you can go in and fix them and clearly see for example, there is a data feed missing from Bloomberg. It's usually something really simple.

Lars Wind: One of the most important things is to check for consistency in data, and that data that you get daily is actually correct. But you can easily set up procedures to do that. I first built the system at ADIA, I was a one-man team, then Betina joined and that required I explain the whole process. There was a lot of cleaning to be done. And just more recently, we all three are able to run the entire quant framework independently.

Simon Hampshire: If it was written in Python, I would never have been able to handle it. It enables the whole human intervention input as the overlay. You are not going to get a Knight Capital occurrence, you are not going to cause a mini-crash from models because there is a sanity check on a daily basis.

Betina Wolf-Andersen: We have a rule that all the trades we have on should make sense. You should be able to explain them to a non-investment person the logic behind it. If you can't, there is probably something wrong. You can have a great model that works, but it has to work in the context of the real world. That is also what we try to have as one of our core focuses and values. Everything we do makes sense, and not just from a technology point of view.