Michael Gleason, head of quantitative strategies (MG): If we go back to 2000, when we were working at Putnam in Boston, Gartmore wanted to do two things in London and globally. They wanted to raise the bar in terms of the quantitative tools and techniques they were using in the business as a whole – that is risk analysis, attribution, back testing, screening and stock selection. Also they wanted to build their market neutral book of business. To achieve both objectives they needed many of the same things: to subscribe to the world's best data sources, to hire information systems specialists, put servers in place and start building the analytics you need to do both of those things.
This was a brief discussion between the firm and myself in 2000; it progressed through 2000 and took until spring of 2001 for me to arrive here. The first six months to a year were spent putting the architecture in place, the next six months to a year were spent building some analytics and core models and hiring some people. In 2003, the team achieved full strength when Luke and Andrew Ver Plank joined from Putnam. We are old colleagues so we spokethe same language, there was very little up time required in terms of a learning curve; we could hit the ground running.
At the end of 2003 we won our first mandate, a long/short market neutral mandate from Lyxor, who have seeded several Gartmore funds. The rest is history.
MG: What we are talking about with market neutral is a risk controlled, portfolio construction process in long/short equity. It is about harvesting information from multiple sources, integrating them into a unified alpha and then building a portfolio with superior risk and return characteristics.
It is unlike anything Gartmore has, or has ever had. There are some competitors out there in the marketplace, but we think we have some unique features here that separate us from about 99% of the competition.
The most recent year has been amazing in terms of growth. We have gone from that one managed account of about $11m up to more than $550m in market neutral hedge funds. We have also got about half a billion in long only funds. So it's been an amazing period of growth. If you think about how we have done this, because our returns were pretty stellar in a year that was pretty lacklustre for most funds, you can really sum it up as people, process and technology.
The people are clearly driving this process; it is not about stat arb or volatility arb, this is about a structured, fundamental approach. We are all working on research, everyone is very hands-on, they are very involved in the process.
That process is key as well. It is a risk controlled portfolio construction process. We're using an optimiser to constrain for all kind of risk: beta, beta within sub sector, dividend yield, FX, country, sector and many others.
The technology is also key, that's our third advantage. We have the best technology, in our opinion, in the business. It is only used by about ten firms in the world and we are the only European client – it is called Vision.
In my view, we're feeding off a cognitive dissonance that exists in the marketplace. The market might be efficient but it's organisationally dysfunctional.
MG: It's a very quiet, but solid, truth that most managers do not perform well. They underperform their long only benchmarks; also most hedge funds do not perform if you take out the market effect because they are only buying beta. So why is it that we are able to perform? We are feeding off their cognitive dissonance: they are trading too much, their processes aren't locked down and are organisationally dysfunctional. This goes for most hedge funds and most long only giants.
We are there to pick up the basis points that they are leaving behind. Our hit rate is only 54%, so we are not right all the time – it's a very slim margin of victory. When you apply some leverage, though, you get to the kinds of returns we are talking about, about 12-15% net with risk of less than half that.
Luke Smith, senior quantitative portfolio manager (LS): The process itself starts with what we call our quantitative factor library. A factor is any variable we can use to rate stocks, a classic example being P/E ratio. A good quant factor can differentiate good stocks from bad stocks over a long period of time.
How powerful a factor is, how consistent it is and what else it is correlated with are all things we look at. We have close to 100 factors we look at to evaluate 10,000 different stocks globally, every single day.
The first step of the process is to tabulate each one of these hundred different ratios or variables for each of the 10,000 securities. The result of that goes back into the database; that's the factor library. When that's done, the quant model runs. The purpose of the model is to convert relative rankings on each of these measures into expected future returns.
The way it works, again thinking about P/E ratio, is if you pick a group of relevant peers to a company and look at their P/E ratios you are going to get something that looks like a bell curve. A company that is, say, one standard deviation above average historically in terms of P/E would historically generate a given level of outperformance. That's based on the historical analysis, so in a sense each one of these factors is giving us a return forecast. The rest of the model is a matter of blending these different forecasts together.
We give different weights to different factors, again based heavily on our historical research. A lot of people do this, but where they go wrong is that they set the weights to maximise performance in their back test period. What we want to do is discover the best performance going forward, and that is where it takes experience and judgement rather than just statistics to fit the historical data. That is where we really think we differentiate ourselves from other people.
Then we go to the portfolio construction stage, which is where we use the optimiser to build the portfolio. The three inputs to the optimiser are the return forecast that we generate from our model, risk models that forecast future volatility and the cost of building these positions: spreads in the markets, borrow rates for short positions and so on.
What we're asking the optimiser to do is construct a portfolio that maximises the ratio of net return to volatility. Everything in there is forecast, so it shows returns minus cost with relation to volatility. That is what is known as the information ratio, and our process was designed to create the highest information ratio possible.
In the optimisation process we can constrain various characteristics of the portfolio. The major ones people will look at are net dollars long versus short – where we are very close to even – the beta of the portfolio, which is almost exactly zero and country bets, of which there are none whatsoever. What you can see is that we have exposure to both momentum and valuation, because those are the major classes of factors in our model.
If we create a portfolio with these characteristics and no sensitivity to anything else it will produce very consistent returns over time.
MG: Yes, we are severely capacity constrained. I think we are probably more aware than most people about the limits we are able to achieve. We have constraints on average daily volume and how many days' volume we are able to hold. We understand transaction costs from beginning to end, so we realise – and it is a very painful truth to recognise this up front – that capacity is not limitless.
LS: If you break down the universe in terms of market cap buckets, biggest to smallest, and have a look at how effective our model is you can see that our annual returns are about 35% in the smaller space and only about 10% in the largest. However, if we are in the smaller end there is much less liquidity and it is going to cost us much more to trade, whereas further up the scale it is very cheap to trade but it is also a much more efficient market.
So we could take in more money, but that would force us to move into the larger cap position, which would take the returns down. So what we are trying to do is keep it fairly evenly balanced across all caps and that does limit us in how much we can take in.
THFJ: Does that create an issue for investors? In order to make it a worthwhile, generative portion of someone's portfolio they might want to put in more money than you can take.
MG: It's still a pretty large size: Regulus holds over $400m. Crucis is smaller, about $150m, and we've closed it. Also, the next fund coming up, our US fund Absolus, has got a capacity of close to a billion. So people are writing very large tickets, and there is room for them. The next funds that will be coming will be a Japan and a global construct and those funds will also be very large.
THFJ: Why the decision, then, to go for Europe and Asia ex-Japan before the US and global funds?
MG: A couple of reasons. On the European fund I think we played to Gartmore's strength: the entire research process here is geared towards UK and European equities. Also, we know the alpha works best in mid- to small-cap non-US markets, so Europe just seemed the best choice. With Asia-Pacific, I think, we were able to bring our technology and process to bear in a very difficult segment. It is also a very unusual product; there are not very many market neutral Asia-Pacific ex-Japan quant funds out there.
LS: If the markets become more liquid, yes. The historical returns for the strategy in Asia are very high. What you will find is that in developed markets the returns are around 20% a year, which is still very attractive, but if you look to Asia ex-Japan and emerging markets they are 35-40% a year.
The problem is that in emerging markets it is very difficult to implement the strategy; you are not allowed to short in a lot of the markets. The Asia ex-Japan space, what we would call 'developed Asia' – Australia, New Zealand, Singapore, and Hong Kong – is in between the developed and emerging worlds. Returns are still pretty high and yet we are able to implement the strategy. However, the liquidity of that market is about one tenth that of the European universe.
THFJ: Is the global portfolio going to be completely separately managed or a synthesis of the others?
LS: The model that drives it is going to be identical, so the portfolio is going to look very similar to what would happen if you put together the four regional portfolios.But it will be managed as a separate identity.
MG: The US fund launches on 1 June, and then it will be late Q3 to mid Q4 before the fourth fund is launched.
LS: The attraction of this is that the upfront investment to get Vision and the team in place here was very large; it is not a cheap system. Also you have to buy all the data from the vendors, and it was a two to three year period before we were bringing in any revenue. But once it was all in place, it is incredibly scalable. We evaluate 10,000 companies from around the globe in an hour every day and then we can run multiple products off that with the same team.
LS: Gartmore provides all the peripheral functions so we can focus on the investment side. They also had the scale to make the investment in the Vision system and to wait two years to get some payback on that. A smaller firm would not be able to do that, and if we were doing it on our own we would have had to go pretty deeply into debt before seeing any revenue opportunities.
MG: Setting up your own shop with no previous hedge fund experience would be near impossible with this strategy. People often ask this, but we clearly needed to get established here at Gartmore and pay our dues by supporting the investment division and building the analytics that they can use. We are sending reports around the world and there are hundreds of users within the firm that just receive the alpha reports.
LS: The strategy is very rational, so when it doesn't work it is when the market is behaving irrationally. We can measure that: when you look at high beta, volatile stocks, when they are performing extremely well our strategy doesn't work. We have tried to make the portfolio neutral to beta and to volatility so in those types of markets our expected return goes to zero. It is no longer positive but it is not negative. That means we are not going to lose money in the type of environment where the model doesn't work. We're not going to make any either, but we have tried to protect ourselves from the downside.
The classic example was in the peak of the bubble in 1999, the model had five or six straight down months by one or two percent; the total drawdown was about 6.5% over the whole period, and then it came back very strongly when the bubble burst. It wouldn't have been very pretty, but we would have survived that period. Compared to some people that would have been a pretty good return, and we were rewarded in the next three years.
That is very difficult to do. A lot of people after six straight down months would have changed the model. The most likely thing to do would have been to put less weight on valuation, because that was getting killed, and put more on momentum. But then when the market turned you would have got hurt the other way as well.
MG: The key element I think is the behavioural side. A lot of what we're doing is very similar to what fundamental managers around the world are doing, but for the alpha to erode you would have to see dramatic changes in the behavioural element of the human psyche, and those things take a long time to change. So as long as we are patient we think this process will continue.
If you look at surveys from the sell side, quant market neutral in general is losing fame. So there are not many new participants coming in here, and the best ones have been closed for years, so we are not too concerned about new entrants popping up with spreadsheets in garages somewhere.
THFJ: Probably more than any other strategy, people turn their noses up at market neutral. What would you say to them?
MG: I think the facts tell a different story. I think that most hedge funds are beta peddlers, and I think that to say market neutral is somehow a second-class citizen is really unfair.
It does depend on the process and the team running it. You can make a tremendous level of money at an acceptable risk level for your clients [in market neutral] and there are lots of natural competitors out there for this process that have been around for many, many years and have been full for many, many years.
LS: My view is if you want market exposure, buy an index fund. That's the best way to do it. I think people were turning their noses up at market neutral in the peak of the bubble when markets were going up 20-30% a year, but you get the downside of that and markets went down for three years. The people who take a long-term view and look at markets over the long run really appreciate the value of the market neutral strategy.
THFJ: Other than your own, what hedge fund strategies do you think can generate alpha – where would you invest?
MG: We believe in process driven approaches, so in that case credit is a natural alternative. The CDS market is very deep and liquid. A credit approach that was risk controlled and process driven would have some promise.
MG: In this environment, and I know it's fashionable, but I would not be touching global macro. There are too few bets; we believe in diversification, that you should not get paid for taking on risk that is diversifiable. Betting on four macro factors doesn't seem sensible to me.
We only need to be right 54% of the time in this process. If you're running with just four to a dozen bets, your hit rate needs to be about 80% to achieve the same returns as we are. You had better be damn good to get a hit rate of 80%.
THFJ: In terms of where you want to go with this strategy, what else can you do with it?
MG: We would be thrilled to achieve full capacity in four or five funds. However, we do have our strategic hats on trying to figure out what else we can use on this scalable platform. Credit/ equity hybrid strategies are possible, and we could look into hybrid long only where you can go short 20-30%.
The regulatory environment is changing, so perhaps there are some things we can do that plug into the periphery of our architecture.We are considering those things, but our first priority is to fill our entire market neutral book.
LS: The expectation is also that markets become more efficient over time, so as we're exploiting inefficiency, taking more money in means we have a lot more work to do to maintain the returns. They will erode away, and we have to find new sources of alpha – that's a lot of work.