Hedge Funds, Have investors got a measure of Quants

By Helen Avery, June 2007

Quantitative hedge funds are increasing in number. Larger ones with the money to invest in research, technology and staff are becoming ever bigger while smaller quant funds struggle to keep up. Are quantitative strategies the sure-fire way to uncover and pin down alpha, as many investors are beginning to believe, or is human intervention in their implementation still all-too important? Helen Avery reports.

"I CALL IT the revenge of the nerds," says Andrew Lo, a professor at Massachusetts Institute of Technology, and director of the university’s Laboratory for Financial Engineering. Lo is referring to the growth in demand for statisticians, mathematicians, physicists and astrophysicists from hedge funds, asset managers and banks. As one hedge fund quant employee advises a student enquiring about a future career in finance on a US website: "Forget taking economics. Stick to mathematical modelling and programming."

It is an interesting transformation. "In the US, Harvard and Yale have tended to be the suppliers to the finance employment market but now universities such as MIT and Caltech that produce technologists are beginning to have an advantage," says Lo. He is bound to be a little biased, but his statement is supported across Wall Street – quantitative investment strategies, often involving advanced computer programming and mathematical modelling, are becoming an increasingly common feature of the financial landscape.

The demand for staff who can develop and monitor complex algorithms in the search for alpha, and the attractive fees it can bring, is probably at its fiercest among hedge funds. Kyle Ramkinssoon is a principal at US-based recruitment consultant IJC Partners. Most of the firm’s clients are hedge funds recruiting for quant-oriented talent. "There is a huge shortage of talent and it is very common for candidates to receive multiple competing offers, as well as counter-offers," Ramkinssoon says. "Our client looks for candidates with degrees in computer science, mathematics, statistics and physics. Some other degree categories are considered, but they have to come from top schools and the applicants must have high GPAs and high standardized test scores.

"On the entry side for the quant developers and quant analysts, there is a lot of programming involved, hence the demand for the IT skills – they develop the models, work with the traders, portfolio managers and other quants to really drill down on these models. They are finding staff by going directly to the schools at the freshmen level and earlier and staying in touch with these stellar candidates. Others are doing on-campus recruitment, tapping their vast alumni network or using multiple recruiters to find them talent.

"There is a tremendous amount of competition as the big funds are looking to add staff at a huge clip.

We’re working with a fund managing over $20 billion and they’re looking to add 43 additional head count – four traders, and the rest quant developers and pure IT development."

The rise of quant

Industry participants point to the increase in the number of equity market neutral strategies being run as evidence of the growth in quantitative hedge funds. It serves as only a rough guide, and can encompass funds using quantitative screening only, up to the high-frequency traders with developed buy/sell models executed by computers. The high-frequency traders that are considered pure quants, such as DE Shaw, Renaissance Technologies and AQR, are estimated to number only about 20 worldwide. But given that nearly all quantitative managers need to be in the equity market neutral space or long/short equity space for liquidity purposes if relying on models, it is a good litmus test.

According to data from HFR, the assets under management of equity market neutral funds increased by almost 50% from the beginning of 2005 to the end of 2006. These figures do not include some of the largest quant funds.

There are several reasons for the increase, at the heart of which lies investor demand. The greatest attraction for investors is that quantitative-based investing can eliminate manager emotion. Lo says: "Investors want a degree of stability, and putting portfolio management into the hands of a process rather than a person can remove some potential errors. For institutional investors with a fiduciary responsibility, there is going to be more comfort with a process than with allocating to an individual."

Investors also appear to be becoming less sceptical about the so-called black-box approach of quantitative strategies. Although they may not fully understand how an algorithm is developed, they can at least understand the principle behind it. Lo says that large quantitative houses have made strides to provide more transparency and to educate the investment community, so that "black boxes are becoming glass boxes". Alex Greyserman, CIO at hedge fund Hite Capital, and an adjunct professor at ColumbiaUniversity, says: "I think people are beginning to see that a manager who is developing or selecting a quant model to pick stocks is no less transparent than a manager that picks stocks based on proprietary research or experience."

The strong performance of quantitative hedge funds has naturally aroused further interest. Sunil Pai, co-founder of ST Capital Partners, launched a quantitative hedge fund in February. "With 70% of the world’s money being invested with humans using fundamental analysis, there is a lot of identifiable alpha out there so long as you can see into the data," he says. "With the tech boom of the last 10 years, there are now machines fast enough to crunch this data which, combined with the right mathematics, can find patterns off which managers can trade using a quantitative approach."

HFR’s equity market neutral index returned 2.7% by the end of April, 7.28% in 2006 and 6.22% in 2005. "We’ve gone through a market cycle that has been uni-directional for some time so the models that have been developed and work have enjoyed continued performance. Also, since the internet boom there have been huge pools of liquidity, which has helped such funds that rely on their ability to trade with high frequency," says Paul Alapat, managing director and head of quantitative services at investment research outsourcing firm Amba Research.

Reportedly the large pure quantitative hedge funds are returning considerably more. Renaissance Technologies, started by maths professor Jim Simons, has produced net annual returns of 37% in its $6 billion Medallion Fund since 1989. DE Shaw, a $26 billion quant house turned multi-strategy fund, has reportedly posted average annual returns of more than 20% since its inception. AQR, a $35 billion quant hedge fund, is said to have also produced average annual returns in the double digits since its establishment in 1998 by Clifford Asness and other members of the Goldman Sachs quantitative research group.

The virtuous circle

The disparity in returns between the large and smaller quantitative hedge funds highlights the challenges in the sector.

Although it stands to reason that smaller funds not forced to put billions of dollars to work are at an advantage, the large funds have the money to scoop up skilled individuals to develop and tweak algorithms, as well as being able to benefit from lower execution costs, allowing even higher-frequency trading.

One hedge fund manager comments: "Today it is not so much about finding a model with an edge, but rather about building an extraordinary research and development team that will constantly identify new anomalies.

"This cost of building and maintaining a world-class team is in the tens to hundreds of millions of dollars a year. This makes success a virtuous cycle. The better you do, the more capital you have to do more research and development, and the more adaptive and successful you become, thus leading to more profits and even better research capabilities. Therefore, the edge of the top firms will continue to get better and create very high barriers to entry."

Lo is not as convinced that the smaller funds will disappear, leaving only the DE Shaws and RenTechs to carry the quantitative torch. "Yes, the competition is increasing," he says, "but there are always innovations occurring, sometimes at levels where big firms will not take an interest. If you are a $10 billion fund you need to generate a lot of ideas to produce 5%, for example."

Nonetheless, a trend is emerging of platforms being set up to seed or help start-up quantitative hedge funds and provide them with the technology and execution capabilities to enable them to compete with their larger peers. James Casper, for example, has set up a "fund of algorithms" called Olive Tree Capital. The due diligence process can be time-consuming, however. Casper and his team have developed a structured methodology that can analyse trading algorithms, understand how they yield alpha and uncover risk factors. These underlying algorithms are then stress tested – a process that can take up to a year. The process analyses as many as 500 algorithms a year, out of which Olive Tree allocates only to a diversified pool of the most successful. Last year, the algorithms were running more than $800 million. Olive Tree launched a public fund at the beginning of May that currently has nine underlying quantitative funds, and a projected capacity of $10 billion in just two years’ time. Olive Tree has an algorithm that consistently monitors the underlying algorithms, looking for anomalies, and checking they are running in sync. "We’re as quant as quant gets," says Casper.

It’s an interesting alternative to setting up shop on one’s own. Casper says as many as 10 algorithms are presented to the firm every week. "These developers often have two choices. They can set up on their own, or they can approach the large quant houses with their models," he says.

The difficulty with the first option is execution and operational costs. It is practically impossible for a small operation to build or acquire the needed risk controls and infrastructure. Most prime brokers and auditors will not even take on a client that has less than $50 million to $100 million under management. The service providers are simply operating at overcapacity and their return on invested time needs to be allocated to the bigger clients.

In addition, Russ Koesterich, a senior portfolio manager in the US market neutral group of BGI, points out: "There are clear benefits to scale. Data needs to be thoroughly cleaned, so having the technology and infrastructure to do that is an advantage. Reducing transaction costs is also crucial. Making smaller bets across many names reduces liquidity risk and reduces costs, but requires more infrastructure." Large firms are also able to reduce execution costs as they are conducting a large volume of trades. If you run $5 million, your broker won’t execute your trade as efficiently as he will if you are running $5 billion. "It’s a bit of a Catch-22," says Greyserman. "Some of these large firms do thousands of trades a day, using hundreds of computers to search for arbitrage opportunities every few seconds. You have to be big to do that in the first place, but then execution and transactions costs are cheaper and you can play opportunities that others cannot afford to play."

Intellectual property issues

Some developers are reluctant, however, to join the large quant firms and become swallowed up in the machine. Casper says: "It is difficult to approach the large institutions. Even if you want to give your algorithm to them for free, it can be difficult finding the right person and department and ensuring they understand what you are doing. When they come to us, we can understand what it means to build a system as we are mathematicians, traders and risk managers. We are able to be more responsive and attentive." There can also be a reluctance on the part of some algorithmic developers to join a large firm and run the risk of losing the intellectual property. Casper’s firm, for example, does not always insist on access to intellectual property and, when it does, the developer is guaranteed full protection. One developer claims he was offered a $150 million allocation to his algorithm by one of the largest quantitative hedge funds, but feared that if he accepted the money he might be kicked out later – his algorithm no longer his property.

Intellectual property is an issue that all quant hedge funds have to contend with. Some large quant shop employees lament the paranoia that seems to plague certain institutions.

One former employee with a European quantitative hedge fund, who has now opted to join a platform, says: "The prospective benefit of working at a big hedge fund is the opportunity to share some of what you know with others and you can pick up some new ideas that may help improve your own trading systems. But the irony is that often these firms prefer to foster internal competitiveness rather than communication.

"Traders tend to sit at their Bloomberg screens like poker players hiding their cards. As a designer of trading algorithms I am seeking to set up trading systems that trade without my constant attention. This leaves time to study and learn other people’s techniques. At a big firm, highly qualified experts are presumably wrestling with many of the same problems. I am willing to share a lot of what I’ve learned on the basis that it’s hard to replicate an algorithm out of context. There are many things I’ve studied that I don’t use and my now-unused analysis may be of some value. If every trader is afraid that their work might be stolen by the trader at the next workstation then each one will be busy reinventing the wheel in their drive to discover the philosopher’s stone."

Small funds in particular face difficulties retaining staff and therefore intellectual property, especially with the huge recruitment drives being undertaken by the large pure quant players. Amba Research set up a group 18 months ago that outsources qualified and trained professionals with backgrounds in engineering, mathematics and statistics to smaller quant funds in need of staff. Amba’s Alapat says: "A lot of information gets transferred to analysts who can shift from one firm to another. We’re seeing a lot of that happening in London and New York. Communication confidentiality is crucial. We have non-disclosure agreements. If there is a dilution of information that is essentially a lost trade for a client." The group already has 65 analysts.

Quant as a panacea

While holding on to intellectual property is crucial to the success of a quant fund, equally important is holding on to investors’ money. "Some of the smaller quant funds are at a disadvantage because they are forced to take in hot money to get going," says François-Serge Lhabitant, associate professor of finance at EDHECBusinessSchool and CIO at Kedge Capital. "Not being able to plan long-term allocation makes business difficult."

Investors’ growing interest in quantitative strategies raises some questions. Investors are naturally tempted by the returns, and feel somewhat securer dealing with scientists and mathematicians and computers than with the gunslinger egos that the hedge fund industry sometimes suffers from.

But risk management is still essential. "Let’s not forget, we have only seen one cycle for most of these funds," says Alapat. "We’ve had five to six years of a stable market. It will be interesting to see what happens to these models if there is an abrupt reversal in market conditions."

Investors can be prone to forget that the models on which these quant funds rely have been developed using historical data, and markets can change. Humans are still in charge of programming and tweaking the models. "In a sense you’re still investing in a discretionary manager," says Greyserman. "You need to make sure that the manager does not get too attached to his model and is willing to adapt and admit if certain models are no longer working or the market has changed, or the opportunities have been arbed away."

The collapse in 1998 of Long Term Capital Management serves as a warning.

Although not a quant fund, it based its investment strategy on a model developed by a team with significant quantitative experience. However, when markets did not react in line with the model’s assumptions, the lack of risk management, and significant leverage in the firm, allowed the losses to reach phenomenal proportions.

"Quantitative strategies should have good downside risk management," says Lhabitant. "How much can you lose and where should you stop? You can look at a stream of numbers all day and lose touch with what is happening in the market." His point is that some degree of financial knowledge or experience is essential in addition to a mathematics background.

He continues: "What if the model is built to sell a company at 20, but there is no buyer? If a market becomes illiquid a computer can’t tell you that. Unless you include liquidity in your models, which some small funds don’t, then the model may not always work. Similarly, if there is a stock that splits, going from 100 to 50, and the model doesn’t know – it will think it is a great buy and execute the trade, but it isn’t a good buy at all."

Lhabitant says that as an investor he is more comfortable with quantitative managers who have lost money in the past, and therefore have learned the effects of changes in market environment. Interestingly, three of the former LTCM team announced in May that they would be setting up a quantitative fund called Quantitative Alternatives.

Lhabitant says: "Our approach is to understand the rationale behind the model, and then the viability of the model over time – what is the goal? And why should it work? Is it being constantly monitored? Is it evolving with the markets? Is it being fine-tuned? And then you have to look at the risk management surrounding that. Let’s say, a manager comes to us and says: ‘There is a bizarre reaction just after auction of treasuries and so we have an algorithm to exploit that’. We can understand the model, but is that manager measuring the risk, what are the sizes of trades? What risk controls are in place? And if he is wrong, how much will he lose? It’s the same for investing in any hedge fund. Don’t put money in unless you really understand the risk you are taking."

The extent to which a fund is quantitative should also be considered. If a fund is purely screening stocks based on a computer model but uses fundamental analysis or has humans making the ultimate say on trading, investors should be clear about this. "There is concern that there are people who are not as quant as they propose to be. The buyer just has to be aware of these possibilities when doing due diligence," says Lo.

The future of quant

Given quantitative investment’s growth over the past five years, is it set to take over from investing based on fundamental analysis as the future of hedge funds? Lo is convinced this is the case. "With advancements in technology, I think fundamental guys unable to think quantitatively will be at a disadvantage," he says.

"Soros and Buffett are exceptions to the rule I believe.

"The technology of the next 10 to 20 years is going to create new products and services in the financial sector, and transparency is going to increase. The question is – where is the money being made coming from? If it is a zero-sum game, technology can only get you so far."

A future of investment decisions being executed by large rooms of computers being programmed by scientists and technologists is perhaps still too like a science-fiction novel to truly comprehend at present.

Lhabitant argues that it is less about the future of technology when it comes to hedge funds as about being able to act outside of the box. "If there are 20 million models trading, then the only way to beat a model might be to use fundamental analysis and human intervention," he says. "I think there is room for the two. It will be a balance." The business graduates of Harvard and Yale heading for Wall Street need not worry too much just yet.