21 Want to Build a Computer That Makes Money? with Nigol Koulajian of Quest Partners - 1of2

Niels

Welcome to another episode of Top Traders Unplugged. Thanks so much for tuning in today I know how valuable your time is, so I appreciate you spending some of it here with me today. Now on today’s show I’m talking to Nigol Koulajian, founder and CIO of Quest Partners. Originally of Armenian background, Nigol spent part of his childhood in Lebanon, but by the age of 16, he moved to the US to escape the wars around him. The uncertainty he had to live through as a child has clearly influenced his approach to life, and later on, the way he designed his trading strategy. In a very personal and detailed conversation, we discuss some of the major risks that Nigol sees in how investors are perceiving skill and alpha with some of the large hedge funds today. What he does to differentiate himself from these risks, and how daily meditation helps him see the world with clarity, which allows him to focus on continuing to build a solution oriented alternative investment firm. And for those of you who are new to the show. I just want to let you know that you can find all of the show notes, including a full transcript of today’s episode on the TOPTRADERSUNPLUGGED.COM web site. Now let’s get on with part 1 of my conversation. I hope you will enjoy it.

Niels

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

Nigol

Thank you, Niels.

Niels

Now, as I was preparing for our conversation today, I noticed a few really interesting things that I’m sure that we’ll have a chance to discuss today, but here are some of my initial thoughts. One is the fact that you have a long, and may I say very solid track record in your original program, but the program itself is a bit different to many of your peers, which I found fascinating. Also, you seen to have diversified your business into other product types, such as an equity program, both hedged and long only, as well as a tracker index where you seek to deliver the returns of your CTA peers, and I would be tempted to say that your firm seems to be looking to find solutions for investors, rather than being a purist in one particular strategy. And finally, I noticed that you actually started on the other side of the table, so to speak, namely as an allocator or investor into hedge funds and CTAs. So that in itself is of course an interesting way into what you do today. So I’m really excited about all these possible topics that we can talk about today, but, of course before we go into too much details about your company and where you are today, I would really like if you could take us all the way back to the beginning, telling us your story, and what led you to take this path. Also feel free to go back as far as you want, and share how you met your business partner, Paul, and I’ll be open here and say I’m going to let you pronounce his surname because I don’t think I’m going to do that justice.

Nigol

It’s Paul Czkwianianc.

Niels

Thank you.

Nigol

It’s a polish name actually. Great, thanks Niels. So in terms of background, I guess I’ll go all the way back to the beginning. I was born in Lebanon of Armenian descent, so I grew up in the Middle Eastern culture, and also within an Armenian family, so a lot of tradition and that sort of thing. This somehow came into play later in terms of the way we actually look at the world, the stability of the world, etcetera, so we’ll go back to this soon enough. At age 16 I came to the States, due to the war in Lebanon and becoming a little bit too intense. I went to college at Notre Dame to study Electrical Engineering. From there I came back to New York and I went into consulting. I worked for Anderson Consulting, and at Anderson I spent most of my time working at Solomon Brothers, totally by chance, and from the engineering background we went to more financial type of application of programming and modeling for mortgage backed securities, pre-payment models, and that sort of thing, completely by chance. So I became more and more interested in finance. Of course I was working in downtown New York, and Wall Street was very impressive at the time. So I started really doing a lot of study on my own. I started programming on my own, started designing models in 1991. From there I decided to get a more formal education in finance, so I went to Columbia Business School. During business school I was mainly spending my time on programming and designing models and all kinds of things like that.

Niels

When you say models, just to put things in context, what kinds of models were they at the very, very beginning?

Nigol

You know, funny enough, way early at the beginning, before looking at trend following, I looked at volatility breakout models. Which means much, much more short term, day trading, holding the trade 3 to 4 days per trade, that sort of thing, and it’s only later on that I actually started developing trend following models. It’s sort of funny, but I used to read a newspaper called Investor’s Business Daily that was giving investors a lot of tools to evaluate different stocks using different measuring techniques, some fundamentals, some technical, and obviously the fundamental factors were difficult to test, but the technicals were pretty easy to test and to program. So I got the data, I started developing a testing platform, and it’s due to that newspaper that I went into technical training and designing models.

Niels

Interesting.

Nigol

So for volatility breakout, actually, it was pretty early on. I guess at the time it was Monroe Trout and eventually Toby Crabel and those sorts of names that started using it. So, during business school I was interacting with all of the professors at Columbia. Columbia is more directed towards equity and fundamental investing and that sort of thing, and I was going around with my models and showing this to professors, and nobody was getting very excited, let’s put it that way (laugh). During business school I also worked at Deutsche Bank and did value at risk for the US Bank over the summer and into the second year. That was also still in the early 1990s and at that time value at risk was relatively novel. So I came out of business school in 1994. I had models ready to go to start the CTA – short term models, long term models, you name it, I had a platform already to keep track of trades, download data, all kinds of things. After business school, rather than quickly going into the industry, I spent another six months still working on developing models. 100% of my time was dedicated to research at the time. Finally, at the end of 1994, by chance, I got in contact with Victor Tasher, who at the time was running a risk arbitrage hedge fund. He was looking for a trader and I guess he was impressed by my background, and he told me, “this, what you are doing, is very interesting. I like it, but I need a trader, so why don’t you join me as a trader and we’ll do your stuff down the line, once there’s time for it.” So from looking at it from a place where I was going to raise money to start the CTA, and the next thing you know I’m trading in a risk arbitrage hedge fund, designing hedges for all kinds of positions all over the world, and trying to reduce overall market risk, and trying to really isolate specific factors within the stocks that we were actually holding. We were trading risk arbitrage globally, and really there was never any time to focus on the CTA strategies. Eventually I moved on, but it was a very, very interesting experience for me because I was so used to trading in the direction of the price action and in this fund we were actually trading very, very fundamentally driven, very value driven, sometimes in liquid positions where we would, based on our buying – I was actually influencing the price, even in a small fund at the time. It was, from a psychological perspective it erased everything that I knew about how to trade, and I realized that there are other ways. It really broadened my horizons pretty substantially.

Niels

Interesting and it sounds like a great experience to get exposed to early on.

Nigol

Definitely, especially because I wasn’t looking for it. I thought I was sure what I needed to do, and it was a straight-line path, and somehow the detour was very useful, I have to say. So after that I joined Weston Capital. Weston Capital, at the time, was a marketing firm, and they were potentially interested in starting a CTA, and to raise money for a start up for the CTA that I was looking to start. I joined with that intent, and as we were preparing, somehow by chance, the opportunity came to start a fund the fund, and so I said, “oh yeah, that shouldn’t take too long.” At the time the industry was much simpler and due diligence was not as sophisticated and in-depth as it is today. So, we started a couple of fund the funds that allocated to all kinds of strategies including CTAs, and I ran those fund the funds for a couple of years until I realized I was getting sidetracked again, and left and started my own firm, initially as a fund the fund, and then in 1999 a partner joined me and we started a CTA called Enterprise Asset Management.

Niels

So a few detours along the way?

Nigol

A few detours, but I would say I am very glad I went through those. Sometimes not getting what you want is the best thing you can get, because in the long term, these things are very, very useful that we don’t even realize we don’t know.

Niels

I couldn’t agree more.

Nigol

So here we are, we’re running a CTA and a fund of funds – 1999, 2000, 2001. In 2001 my partner and I decided to split, and I started Quest. Paul had joined me in 1999 at Enterprise, and we came in and he decided to follow me here at Quest. We hired the head trader of AHL at the time, Neil Hanover, and off we go (laugh). So our assets rose. At Enterprise we started with about 2 or 3 million of private capital, and at Quest we started with about 25 million, and our assets slowly grew to the first peak, which was around 650 million in assets by around 2007. In 2003 we actually signed an agreement to give 90% capacity rights to all our strategies, to the fund the fund of a large very well-known CTA, and they allocated about 550 million to us at the peak, and we basically just ran money for then from 2003 to 2010.

Niels

Wow, interesting and very unusual, but very interesting indeed. Maybe you can, as a lead on to that…I know we’re going to talk a lot more about the details, but let’s just take a jump and say, so how does the business look today? What kinds of programs do you run, and, roughly, what is the AUM as well?

Nigol

Today we’re managing about 760 million. Of that there’s about 80 million in our traditional CTA strategies, then the replicator has about 80 million as well. The rest is in hedge product – equity hedge and fixed income hedge. I think equity hedge is around 450 million, and fixed income hedge is around 150 million.

Niels

OK, and in total, how many different strategies or programs do you run?

Nigol

We’re offering the original program, which is the benchmark CTA strategy, the QTR which is the replicator, equity hedge, fixed income hedge, and equity long, so I guess that’s six different strategies. If you noticed, most of the assets that we have today are actually in the new strategies, which were out of necessity, we had to develop them because in 2010 that large fund the fund had to redeem due to their own complications. We did very well for them. But as they redeemed them we are in an environment where people are de-allocating from CTAs and we said, OK we need to raise money quickly and what can we do, and we realized it was a question of fees, a question of the relationship between the CTAs and the equity markets were not as clearly defined as they could be, etcetera, etcetera, so we defined these different programs along the way. But we’ll get into that.

Niels

Definitely. It’s super interesting, actually. Now, normally I would go on and ask you a little bit about your business and how it’s structured, but before we do that, today I would like to go in a little bit of a different route. I want to talk more about the broad question, which I think is really important to many investors and perhaps managers may not be focusing on this enough, and that’s really what people, when they look at track records, may be perceiving as the value that they get when they look at a manager. But in reality, it may not be anything to do with skill or alpha. Some people might say that we’ve seen a lot of style drift, because deep down it’s often rooted in the managers that over time change their strategy and their style. I think maybe the last few years we’ve seen quite a big shift in that sense. I’d love to hear your opinion about this and where you see the dangers may lie of an increased correlation between the so called alternative strategies that are designed to protect investors against traditional investment when they have trouble, and the traditional asset classes themselves.

Nigol

Sure, great question, actually. I think that there are some pretty interesting angles on this, especially looking at it from a CTA perspective. So, first I will cover the CTA industry and then I’ll take a look more broadly within the hedge fund universe. So today the CTA industry can be replicated with a very simple model such as 10 day to 100 day simple moving average cross over strategy. So the most basic strategy you can imagine, you apply it equally to four sectors, without any other optimization, you already have something which is 70% correlated to the CTA industry, which is outperforming 3% to 5% a year due to fees. But 10 years ago the CTA industry was looked at a big pool of skill based returns, today it is more and more apparent that at least 90% of those returns are really very easily replicatable, and the techniques of replication are broadly available – so step number 1. Starting with that, saying that the industry can actually be replicated so easily, now if you start looking at the off shoots of this industry – so if you look at the managers that have raised money in the recent years and have grown into the 20 billion, and 10 billion, and 5 billion, etcetera, etcetera, you will see that they actually have exposure to certain factors, and most commonly the factors that they are exposed to have been risk on strategies, so strategies that are correlated to the equity markets one way or another.

So over the years we wrote three research pieces that actually explore – not looking at the CTA industry as one broad basket, but saying specifically what are the factor drifts that could actually affect the returns of CTAs? So I will list those factors and I will go over them quickly and give you an opportunity to see what kind of impact these factors can have. Suppose you are investing in the S&P 500 and you see a mutual fund which is up, let’s say 20% more than the S&P. You say wow, the manager is a super star, we say well, he’s actually invested in micro caps and they have done very well this year. At this stage, considering the transparency has changed the view of the CTA industry, it’s important for the investors to go one step beyond and become aware of the sub factors and their impact. Because the sub factors, that have been performing very well, have very different characteristics during equity corrections, as we will cover.

So let’s look at them quickly. I would say the first factor is the sector optimization, which means CTAs have been allocating more and more to fixed income because fixed income has performed very, very well in the last 20 years. Now it happens to be that fixed income is also the most liquid sector and therefore, CTAs as they are growing, even if it was not their research driven optimization, the allocation issue and the liquidity issue forced them to allocate more to fixed income. So allocating more to fixed income could actually improve your sharp ratio over the last 15 years by 40%. So if you look at a basic model the 10/100, which we will take as the benchmark for the CTA industry, and you increase the fixed income sector allocation, your sharp ratio would go up by 40% which is already very, very substantial. So that’s factor number 1. Now, as a side note, fixed income has been negatively correlated to stocks, which means during equity corrections it’s provided substantial returns to CTAs. I would say the majority of returns that CTAs have generated in the last 15 years during equity corrections, has come as a result of their fixed income exposure. So that’s factor number 1, so sector optimization and there’s a potential for substantial improvement in sharp ratio due to an increase in allocation to fixed income.

Factor number 2 is long versus short. If you look at the returns of CTAs over the last 20 years, you will see that over 90% of the returns of the basic CTA strategy come from the long side, being long trades. Short trades have made almost no money. There were a couple of periods around 1994, 2001 to 2003 very small returns, and then 2007 to 2009 some pretty good returns, but overall, shorts are almost flat. Now effectively this means that the CTA industry is not so different than running a buy and hold strategy on the same basket or the same portfolio that CTAs trade. This is very surprising to most. You’re saying all these trading models, etcetera, etcetera are really minutely different than the buy and hold on the same markets. So we’re saying that the longs, if you traded longs only instead of trading longs and shorts, you would have improved your sharp ratio by 80% over the last 15 years. Now, again, assume you are a large CTA and here you have higher transaction costs due to your size, and shorts are basically at this stage a straight line down due to the higher transaction costs, you would say, “why am I trading shorts? Let me focus more and more on the longs.” So, again, as you grow in size, this factor is something that is imposed on you, and it’s improved your sharp issue by 80% so we see again the potential for self-reinforcing feedback loop where the managers that are growing in size are forced to allocate to factors which have done very well, and I would argue completely by chance. So the third factor is the timeframe. Over the years, and we’ve shown this in our research pieces, over the years CTAs have had to increase the timeframe that they use for their models. Now you can use a 10/100…it used to be in the 1980s and 1990s that 10 day to 40 day simple moving average was the classical trend following strategy, and eventually in the first decade of the new century, it became 10/100 and today, if you look at it, if you do a factor analysis, you will see that the CTAs are trading timeframes as long as 10 day to 500 day moving average, so much more longer term. Here we’re talking where the average days per trade used to be 10, 15, now we’re talking 100 days, 150 days per trade, if you don’t count for rolls. Now, just to give you an idea of how valuable it would have been to trade the 10/500, to begin with the equity curve of 10/500 has not even flat-lined in the last 5 years, such as the rest of the CTA industry. So you look at 10/100, or 10/40 you see since 2009 those models have not made any money. 10/500 is in straight line up without any interruption. So if you optimize the portfolio and add 10/500, you would have the potential to improve your sharp ratio by 70%, instead of .5, we’re talking .8 that sort of level, which takes you to the superstar level immediately. So these are already three factors which have improved your sharp issue by 40%, 80%, and 70% and there’s no skill involved yet.

So a third factor that we talk about in our research pieces is the fact that CTAs make money during trends, but the biggest factor that explains their returns and the value added aspect of the returns is the fact that they make money when the volatility is expanding. So when you put on a trade – a trend following trade, you are making money because of the trend, but you’re really benefitting mainly when the volatility is expanding as the trend starts. Once the trend is established, your returns start to correlate to equities and they start to correlate to fixed income. There’s no more alpha relative to the traditional portfolios. So now, over the years, we show that CTAs…this character of benefiting from the volatility expansion is called positive skew. Basically when you have a surprise it tends to be positive on the returns, and the opposite negative skew is when there is a surprise, it tends to be more negative. Now CTAs over the years in order to generate higher sharp ratios, have taken profits faster and bought the corrections within the trends more and more. First, this improves your sharp ratio, but also it makes you less capable of benefiting during volatility expansions.

Niels

Can you explain that a little, just take a few steps back and then explain that again. I’m not sure I fully understood what you meant by they bought the drawdown.

Nigol

OK, so let’s say you’re running a simple trend following models such as 10/100 and you size your position based on volatility. Now the most profitable trades for CTAs are the trades that were entered when the volatility was low and then expanded once the trade was in place. So now the danger of that is that after a trend is well established, the volatility could have doubled versus when you put the trade on, and now your risk in that specific market is twice what it used to be from a VAR perspective.

Niels

All things being equal, that you haven’t moved your stop and so on and so forth?

Nigol

The stops for the typical trend following models are pretty far away. After a trend they could be two weeks, three weeks away from you. So if you replicate the industry, I would say stops are irrelevant, you don’t have to have stops in the market. You have trailing stops, if the moving average is crossed you trade, but you can trade market on close, you can trade once a week, you can trade once a month as a matter of fact, and your performance would not be very different. So what we’re saying is trend followers classically used to benefit from the acceleration, so prices moving 1% a day and then the trend is established now, this market is like a really hot market, and now that market is moving 3% a day. The typical CTA nor classically you are going to say this is a real trend, and I’m going to be exposed, and I’m going to have a much higher exposure to this market. The volatility has expanded and typically the volatility expands when equities are going down. So when the volatility is expanding, CTAs are expected to generate higher returns.

Now let’s say that CTAs decided that they don’t want to risk as much when the volatility is expanding and they want to make money more consistently. What they would do is they would reduce their position size when the volatility of the markets expanded within the trade, and when the market is…so you are in an up-trend, the volatility expanded slightly, you are going to reduce your position, and when the volatility compresses, which means during a correction, they would add to their positions. So they’re in an uptrend picking bottoms and selling rallies. The benefit of that is that they now have a constant volatility portfolio. They look at their VAR and they are expecting 15% annual volatility every day. The down side of this is that you’ve lost your positive skew; you’ve lost your ability to generate high returns during equity correction.

So factor number 4, and I’m calling it just mean reversion, broadly. Mean reversion means you’re selling rallies within the trend and buying the dips within the trend. Now I want to take this one level beyond and go into the specific example of buying the dips and the financials. It means you can very easily use a model which buys the S&P, if it’s down 3 days in a row, because you’re expecting a rally. Same thing in fixed income. Now we can potentially make this model available for listeners to the podcast, but just to give you a sense of this. If you bought the S&P, if it’s down 3 days in a row, and you exited when you…sorry it’s a little technical but…

Niels

No no, that’s fine, I’ll stop you if it gets too technical, but we’ll take it step by step.

Nigol

It’s a pretty simple model. So you buy the S&P on the 3rd down close, you take a profit if it makes 1/2 the daily range, so you bought it at the close, if it goes up 1/2  the daily range you take profits.

Niels

So the average true range over what period are you looking at?

Nigol

Over 50 days let’s say, not very relevant, but let’s say over 50 days. And you have a stop loss two daily ranges away, this model, over the last 15 years has had a sharp ratio of about 1.2. Even during equity corrections, this model…you can’t even tell from 2007 to 2008 happened. So there’s, in today’s world, which is driven by central banks providing liquidity because markets are driving the economy rather than the other way around, there’s been a preponderance of mean reverting strategies in financial markets with, again I’m say sharp ratios over 1, and without any skill. This is a model that I will provide to the listeners, and CTAs are more and more including such strategies, which are buying the dips in financials as a way to generate alpha. So again, the numbers are really wild, but this type of strategy improves the sharp ratio of the typical trend following model by 130%. It’s very, very substantial. So I’m just giving you the potential things that CTAs are introducing into the portfolios are substantially changing their characteristics, with no skill, which investors have to be aware of if they really want to pick up a CTA with real skill.

Niels

No, it’s very, very important.

Nigol

There are three more factors that we say CTAs have introduced; one is fixed long equity exposure. So if you look at the rolling correlation of the BTOP [Barclay Top 50 Index] which are 50%, the largest CTAs that control 50% of the assets of the CTA industry. You pick out, or you take out the trend following components through a factor of the composition. The residual returns, rolling correlation to the S&P is 70%, so we’re not saying they’re trading equities and their making money on equities going up, we’re saying that they have fixed long positions in equities, that’s very different. Now adding a long position in the S&P just to keep it simple, over the last 15 years would have improved your sharp ratio by 20%, and again, going from a sharp ratio of .5 to .6 takes you from average to almost superstar. At .7 your almost superstar level. So this is how dramatic these factors are. Now the same factor if you short the VIX, so instead of going long with S&P you can short the VIX, short volatility, which has the same effect as being short puts on the buying. Then you could have actually improved your sharp ratio by 90%. Then you have the typical carry models, which improve your sharp ratio by 18%, if you optimize them, and now today your CTAs are adding credit type strategies by exposure to credit swaps and that sort of thing, and those would improve your sharp ratio by about 80%.

So here we’re saying there are seven factors which are absolutely not skill based, which are highly correlated to the BTOP, which means over time you can see that the BTOP is steadily increasing its correlation to those seven factors. And all seven of those factors are things which reduce the ability of CTAs to hedge equity corrections. So CTAs are getting better and better by introducing these factors on a standalone basis, but they are becoming worse and worse at hedging equity corrections. And investors haven’t picked up those factors out of the CTA returns yet, and they are there for focusing on the CTAs that have the most of these factors.

Niels

Sure. Very, very interesting. Incredibly useful, and very, very insightful and people should really pay attention, even though it might sound very technical, but this is important stuff, but let me…I have the same observations. Maybe not so eloquently described as you have done, but I agree with the overall conclusion of what you are saying. The question here, and this is interesting, and it’s particularly interesting and very topical because David Harding, from Winton, was on CNBC, I think yesterday, or the day before, in an interview where he was obviously being asked about equities and so on and so forth, and he kept stressing the point that they may not be a hedge if equities go down, they may be, but they may not be. And so from the way you describe things, I sense that you, and correct me if I am wrong, that you are saying that a lot of these managers that are doing these new things are doing it to compensate for size. Would that be a fair statement, or do you think they’re doing it not really because of size, necessarily, but simply because of trying to introduce them to more stability in their returns?

Nigol

I would say both, and the fact that they don’t have to make a choice between one and the other is even a stronger, or an accelerator for these optimizations. So size – definitely you would have to make these optimizations, but also these optimizations have worked extremely well and are working better and better as time goes on. In the last five years, shorting the VIX has a sharp ratio of again over 1 just a simple, simple strategy. It’s numerical and its size.

Niels

Yeah. Obviously I would love to have David Harding come on the podcast and discuss this issue instead of us trying to discuss it, because obviously there is some reference to people like Winton who have been very, very successful, stabilized the return, reduced the volatility, and now are coming out saying that they may not be a hedge when equity markets go down, and so on and so forth, but let me be devil’s advocate here. Even though I agree with what you are saying, and that is we say it is not due to skill, but what if the skill is that they actually introduced these strategies at a time where they would be beneficial for their returns, but it’s not to say that they won’t deleverage or de-emphasize these strategies later on at a time where maybe the trend following strategies should have more weight. Because we know when we have to accept perhaps with your firm is one of the few exceptions, but we have to accept that for trend following strategies, it’s been a really difficult and tough environment, and that obviously goes back to the fact that volatility has been decreasing and generally trend following strategies make money from the expansion of volatility.

So, I just wonder. Some of these firms, they have 150 PhDs and they have to do something, and maybe they actually came up with this observation saying, “listen, we should increase risk exposure to these types of strategies, because as long as the environment is as it is, and we’re not detecting any changes in the data, they will help?” I don’t know, I’m just putting it out there.

Nigol

Great question, Niels. Let me give you my angle on it. So first, broadly speaking, I would say it’s been… basically predicting factor turnover has been something that I want to say less than 1% of fund managers have been able to do. It’s not something which is broadly done. Nobody, even the macro guys – the Louis Bacon, etcetera, etcetera, at the time where really truly exceptional managers, today cannot tell you where the returns are going to be. If you can actually predict what sector or where the returns are going to be your sharp ratio would be 3 and above. Now going back to the CTAs, so you’re saying is it possible that these CTAs have found a way to tell you when you should exit risk on trading, going to risk off trading, well I would say it’s very highly unlikely. The reason I say that is because we have had an example already of an equity correction after these factors were introduced in those CTAs. So these factors really started peaking into the large CTAs around 2003, 2004, 2005, when the volatility when really low, interest rates went really low, and the carry trades became highly contributory to a CTA portfolio, and then here comes 2007 to 2009, these were able to get out of the risk on trading and go into trend following at the right time. If you look at the returns of these large CTAs or the BTOP 2007 to 2009, relative to the replicator that has a beta of 1 to them. So you replicate the BTOP with a certain…with this beta etcetera. You see that 2007 to 2009 the replicator over the 2 year period made about 96% in return, where the BTOP was up about 17%, so I’m saying that these CTAs that introduced these new styles into their portfolio underperformed the replicator by 80% in absolute return over a three year period.

Niels

Same volatility between the two?

Nigol

Beta of 1, absolutely. So if you look at the way the BTOP had performed relative to the replicator before it has about 75% correlation, and they’re completely in line. So I would say that there already was one major test and they failed, majorly. At a period which was really critical. Now they’re telling you, we’re not going to give you a hedge, we don’t really care, now today we’re a hedge fund, we’re not a CTA any more. That’s a fair choice, but investors should know what they are getting.

Niels

No I agree. As I said, I think it’s really super important and something I probably think is something that you have not only thought a lot about, clearly, but also has shaped the way that you do things, which I’m sure we’ll get to very shortly. But I want to go back and talk a little bit about these six strategies. You have almost 800 million dollars under management. How big a team does it take to run a business like that and how is it structured, and how do you balance between what to do in house and what to do by the means of outsourced providers?

Nigol

There’s one point I’d like to make about those factors, and it’s that the hedge fund industry…a lot of these factors can be understood correctly within the CTA, and within the hedge fund space if people measure risk by measuring tail risk rather than by measuring volatility, or beta to the stock market. So most hedge funds, and most CTAs today are generating numerical alpha to their benchmarks, but the ones that are generating numerical alpha are doing so because they are taking more tail risk than their peers, not because of skill, and that’s something that we went into in our second research piece. And CTAs are doing the same thing. They’re converting positive skew or negative skew with higher sharp ratio. This is something very, very important. So alpha does not equal skill, alpha equals skill sometimes (laugh). So going back to your question in terms of how can you run six strategies, what kind of infrastructure is needed? I would say the most important thing to run, for a firm like ours, infrastructure is needed, but the most important thing is to have a clarity of direction and clarity of purpose. If you are trying to create a product that does everything, the infrastructure is never going to be enough. So the most important thing is to start with a clear mind in terms of what you are trying to offer and to stick to that no matter what the market conditions are.

Now infrastructure wise, because what we do is based on automation, so we’re saying as a starting premise for CTAs we’re saying the human mind is not equipped to make ideal decisions in the financial markets, because the financial market are there to compensate for what feels good. So basically you put it on a trade that feels good, you typically lose money, and in a way you have to be detached from what feels good, and what everybody is thinking and you know you have to be contrarian, and you have to put on a trade that nobody wants and the human mind is not designed to make such decisions. It’s designed to think as part of the crowd. So now we’re saying we’re going to rely on models, we’re going to decide, we’re going to program the models and tell them what to do, but then we have to trust them 100%.

Now that gives you, when you understand how distant you need to be from the models, it gives you the ability to run with a relatively small team, a lot of different strategies, because once they are programmed, as long as they are not over optimized, you can give them life and let them live. You don’t have to be thinking about them every day, whether I need to re-optimize the factors, etcetera, etcetera. So it’s a question of that you have a certain degree of mental space, we believe that mental space is most well spent during the research and the design of the strategy, but then if the strategy is over optimized and it’s going to take too much mental space to maintain, and therefore that’s not a place that we go.

Now infrastructure wise, today, more and more things are becoming commoditized. Whether it’s the accounting, the execution etcetera, etcetera. Now for us we’re pretty good at programming, etcetera, etcetera, so we’ve actually kept in house the order management system, the order execution system. Obviously the trading strategies, we do everything in house, so we don’t outsource. So it’s a lot of the admin, the back office and all that can easily be outsourced and today it’s a commodity and the price spread between different providers is really minimal. So I would say it’s like a no brainer. Over time, now that replication is coming into play, I would say that you are going to see that some strategies are going to become commodity as well. You are going to have five different firms offering what used to be…let’s say a small cap mutual fund in the CTA world is going to be the soft strategies are going to become available exactly the same way through different providers, etcetera, etcetera. The industry is very mature compared to where it used to be 20 years ago, and 10 years ago, and 5 years ago even. With 9 people we are able to…I would say that out of the infrastructure we have, we can service clients, we don’t have that much of an infrastructure in terms of on the asset raising side, because we focused on servicing one client for seven years. Today we have about 5 clients that account for 99% of our assets. So we’re much more focused on client servicing, reporting and that sort of thing rather than asset raising, and as such I would say that our infrastructure is pretty tight.

Niels

Sure, sure, but also still it allows you still to grow as long as you keep that focus, because as you say automation takes care of the day to day stuff, so very interesting. Now looking at the track record, and I think we’ll probably focus, if I may, focus on the original program. We may talk a little bit about some of the other stuff, but if we focus on the original program, because it has the longest track record, going back to May 1999, when you started, and all the way through to now, when people now a days look at track records, they tend to believe that OK, this is what I’m going to get going forward. But that’s obviously not true because most programs go through evolutions and changes over time. But how would you say people should look at your track record when they look at it? Is it very different? Does the program look very different today than when you started? And when were these big changes, when did they occur?

Nigol

So here, again, Niels, what’s really important is to know relative to the benchmark, what bets we’ve made, what factors we’ve exposed ourselves to over time, and whether our philosophy has changed over time. Nobody can predict returns, but if they can know the sub factors that they are exposed to, typically they are pretty happy because they can build a portfolio in the way that they want, and I think that is really critical. So what we’ve done over the years in the original program, we’ve had the philosophy of we’re looking first to be a trend follower – means we’re looking to correlate to the CTA indices, but we also want to generate a lot of alpha, so we’ve generated about 7% of annual alpha to the BTOP per year since inception. Now philosophically, that philosophy has not changed, so our alpha is there, so we’re not looking to be a replicator, we are looking to generate alpha, and we want our alpha to come from positively skewed sources of returns. Which means we don’t want to be bottom picking equities, with a sharp edge of 1.2. We don’t want to be allocating more to fixed income, we don’t want to be going long equities, we don’t want to be shorting VIX, we don’t want any carry, we don’t want any credit spreads. So now how those concepts have been expressed over time has evolved substantially.

So broadly speaking, something which is really important and critical for the firm’s survival is that today 9 sources of alpha out of 10 are negatively skewed. It means they involve mean reversion, they involved being exposed to risk on top of trades, and we disqualify those immediately at the forefront. So now you are left with very few research ideas, but those are typically things that are more difficult to exploit due to transaction costs, but we’re pretty good at that, at controlling that, and that’s how we’ve generated this alpha. Now the benefit of positively skewed sources of alpha is that they are more stable when the market regime changes. So if you tell me, “I have a strategy which is going to buy the S&P when it is down 3 days in a row because the S&P never goes down even two days in a row now.” It’s great, but if there’s any slight shift in market regime, then it’s a very, very large vulnerability. So the track record of the original program, the philosophy has not changed, but the way we’ve expressed this long volatility or long tails, top of bias that we have, has changed over time, and I’ll give you examples.

First, basic strategy – our days per trade is about 7 days per trade, where the typical CTA today is probably around 30 days per trade. The reason we pick the shorter time frames is because that is where we can generate alpha and that’s where you have the most positive skew. So if you want to benefit from tail events, or large increases in volatility, 7 days per trade on average is where you have the most expansion in the volatility. When you go to 30 days, and now people are going to 60 days. So the larger CTAs are of course trading more long term, the more long term you go the less positive skew you generate. So we’re looking for volatility expansion and volatility expansion happens when you cannot predict when, but you can predict the size of it based on its long term average. So typically you would look at the volatility of the market in the recent past, and compare it to a longer term past, and if the recent volatility is much lower than the long term volatility, you would say there’s been some sort of volatility compression. Now from this type of environment, trend following signal, in many different time frames, is going to give you much more positive skew and much more alpha, than what’s typical in the industry which is to instead buy the dips, the volatility has expanded, let me put on the mean reversion trade. So we’re still looking for the volatility compression, although the volatility keeps going down, and it’s becoming more and more difficult because of all of the central banks constantly providing liquidity, but even in the low volatility environment we’re able to generate returns because the volatility has spiked.

So the markets are pushed very far away from equilibrium, and they’re there with very low volatility, but when something goes wrong then the volatility expands much more than it used to. So even last year, when the volatility was extremely low and coming down, we were able to be up in the original program 16% or so because there were four small spikes in volatility…if you look at the VIX for example, and in those 4 periods, that’s when we made about 4% in each, and it ended up 16% on the year. Now, the track record over time…volatility expansion is measured very differently today than what it used to be when we started. When we started you could look at the 4 day volatility compared to 50 day volatility, and if the 4 day volatility was less than the 50 day you would say OK I’m going to take whatever – this signal or that signal or that sort of thing.

Today we’re looking at things in a much more complex way. For example, we measure volatility differently when the market is making new lows and new highs, then when it’s in the middle of the range. There are a lot of people who are protecting price levels, mainly central banks, and other option sellers and that sort of thing, so when a central bank sees for example, dollar/yen go below a certain level, it’s going to intervene and it wants to intervene in a way that minimizes it’s cost, which means that it has to trade as much as possible as quickly as possible to create a dramatic reversal. So if you measure the volatility at the tails, independently of the volatility in the middle of the range, you have a lot of information about the character of the market. That’s an example of going from a basic volatility measure to something which is much more pattern based, and much more relevant in today’s world.

Niels

Would you say that volatility is almost more important for you than price in identifying trends?

Nigol

Absolutely. I would say the typical approach it seems from a replication in today’s world you have a trend following model – everybody starts with that, and they say, “how can I improve it?” Typically what they do is they want to pick the trades that have the highest sharp ratio. As you do that you are looking for those clean trends, those clean trends don’t have volatility expansion in them, and they’re highly correlated to the stock market, they are highly correlated to fixed income. For us, we’re not looking for a trend, we’re looking for the correct volatility setup. We’re looking to generate alpha, and it needs to happen very quickly in time, and that’s only available if the volatility accelerates. So yes, volatility is the most important criteria behind our setup.

Niels

OK, very, very interesting. Now tell me a little bit about how you then constructed the original programs in terms of the kinds of models that you use, and maybe a little bit about (I’m sure they’re not all the same timeframe), how have you, from an overall point of view, constructed the program?

Nigol

So in terms of where we are today, we’re trading anywhere from…all the way from day trades to some long term models that can stay in a trade forever. About 75% to 80% of what we do is time constraint, which means after a certain amount of time we exit the trade no matter what, because the alpha is not there anymore and we expect other CTAs to be there, and therefore we say there’s no point in being here. The trade is not as valuable. So the concepts that we use or utilize across multiple time frames, the key is measuring volatility, but volatility is not the right term, really.

Markets have memory, it means you look at a trend following trades and this is not a random walk. There’s a serial correlation, which is way over 0 and then the market stops from the trend and it goes into a negative serial correlation type of environment. So it’s not that the market overall shows 0 auto correlation day to day, but it’s actually positive sometimes and negative sometimes. Now what is the impact of that on volatility? If you are measuring volatility in a trend, from a VAR perspective, the volatility can go to 0. If you go to the credit market where you are getting paid a certain amount of money every day, the volatility is 0. Now the risk is definitely not 0. OK, so conceptually, our philosophy is that investors are not pricing tail risk correctly because they are using volatility as a measure of risk, and they’re over allocating to markets that have extreme tail risk, being confused, thinking that they’re getting a skill. So, we want to maximize the exposure to that. That exposure happens in the short term timeframes. For example, by looking at the volatility at certain times of the day, which is highly predictable in terms of the future.

Niels

Let’s take a concrete example. Let’s, if you can, let’s just try and visualize it really for the listeners.

Nigol

So I will go back to a model which says compare the 10 day range to the 50 day range. When the 10 day range is small in size relative to the 50 day range then take trend following signals. That’s one example. It’s not really the range, we’re looking at volatility from a statistical way. Volatility should be normal, and when there’s something which is missing in that normality, we’re looking for it to come back with a vengeance for the market to become stable again. If the market only goes up, you would expect a sharp correction. If it goes up without noise you would expect the correction to be much larger than if it goes up with noise. So you’re looking for a pattern of volatility to be symmetrical and healthy. I gave you the example from an absolute level – 10 day range versus 50 day range, but instead of looking at the range, we’re looking at the peculiar characteristics of volatility. I gave you the example of the volatility at the ranges versus the volatility in the middle of the range. Any time you have big differentials in different aspects of volatility, there’s going to be something which is going to rebalance the markets.

Niels

So if I am to understand you correctly, what you are saying is that first you have some kind of volatility filter that determines what kind of model you want to trade?

Nigol

You can replicate with moving averages. Typically most of our trades are based on stops, so again, it’s not channel break out, but it’s based on stops and it’s different than…

Niels

Because you need a trigger…I mean it’s fine to say that if one volatility level is below another then do trend following, but you still need the trend following trigger somehow.

Nigol

So the trigger…you’re going to be OK if you use moving averages, you are going to be OK if you use channel breakouts, so why don’t we…for you to have a complete model, I would say only trade when the volatility has compressed, then trade 10 day channel breakout – Finish. That’s one model. Now the way we do it, we’re using different entry levels, than the channel breakout, but this would give you something which is very different than what the market is doing already.

Niels

And you would add a time stop to that trend following model, even though it’s trend following, and usually trend following implies that you let the trade run. You would actually apply a time stop as well?

Nigol

Absolutely. Because we are looking to maximize the amount of alpha per unit of beta. Beta is easier to get, so we don’t have an issue with that, so we’re looking to maximize the alpha…we say, depending on the setups, so let’s say we’re looking for a volatility compression where we’re looking for the market to correct within a long term up trend, and the volatility to compress, expecting a continuation of the trend with a volatility expansion. That’s the type of trade where you are going to generate some alpha for 2 or 3 days, and very quickly trend followers are going to catch up with you. So you have to set exit for after three days for example. Now on the other side, let’s say the market is in an up-trend and the volatility has acted in a very…there’s an extreme amount of randomness in the market. So randomness…let’s say the serial correlation of the market has gone very negative, after a big up-trend, and now you’re expecting that if a correction comes, it’s going to be very aggressive. So now that type of market, if that correction happens, you can stay in a trade much longer, because it is going to take trend followers two weeks to catch up with you. Therefore, for two week you’re generating alpha. So every set up condition has embedded in it a time after which the alpha switches to beta.

Niels

Do you trade these same…I guess I have many questions, actually, it’s very, very interesting. Firstly I wanted to ask you, how many different models would you say that you run, and does each model get applied to all the markets in the portfolio?

Nigol

That’s very, very important, yes. Based on the stability of market regimes in today’s world, within the CTA space, I would say that not applying models to all markets is a guaranteed over optimization, within the CTA space, in today’s market regime. So, I believe that…OK…you have to be really careful about designing models that work on an individual markets, and you have to say a lot of prayers when you use them. So, yes, we apply models to everything in the portfolio. Now in terms of how we built the original program, we have about 5 different concepts which are applied in two or more different time frames. They’re exactly the same concept. One could be the skew of the market. One could be the volatility compression. One could be the volatility at the extremes versus the middle of the day, that sort of thing. So it’s the same concept, the entry conditions and the timeframes are different, and these things are concepts which survive across markets, and survive across timeframes as well.

Niels

Fascinating. Now this is complex stuff. For you to explain this…a lot of people will have difficulties in getting their head around these types of concepts, because it’s a little bit more than traditional trend following. You buy when the prices move up, and you sell when the prices move down and so on, and so forth. But what I wanted to ask you, because you mentioned in the beginning of our conversation that your upbringing in Lebanon actually influenced the way you designed your systems. So now I want to try and bridge that gap between these very complex models that you just described, and in your upbringing and in inherently unstable environments how does that work?

Nigol

Thanks for the interesting question. Well, you see, we see the world based on our memory of what it’s supposed to be like. It means most people see the world based on their own filters of reality, not based on how it is. So I have filters…

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 guest’s line 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.