Seven years ago David Leinweber, one of the principals of the Californian investment boutique First Quadrant, wrote a paper which posed the question: ‘Is quantitative investing dead?' It was a reasonable question at the end of the 1990s, when there was deep scepticism about the value of quantitative methods.

Yet since then quant investing has clawed its way back into favour. The most recent evidence for this is a survey by Frank Fabozzi, professor of finance at Yale University, and the Paris-based consulting firm Intertek. The survey, ‘Trends in equity portfolio modeling', is a follow-up to Intertek's 2003 survey ‘Trends in quantitative methods in asset management.'

Managers at 38 asset management firms - 23 of them from Europe - with a total of €3.3trn assets under management, took part in the 2006 survey.

Almost a third (30%) of the asset managers say that more than three quarters of their assets are being managed quantitatively. A large majority (84%) report that the percentage of equity assets under quantitative management has increased compared with 2004-05 or has remained stable at about 100% of equity assets.

But perhaps the most significant difference between the 2003 and 2006 surveys is that quantitative methods are now being used widely in active equity management. What was perceived in the past as a mechanistic instrument for passive management is now accepted as a forensic instrument in the search for alpha.

Sergio Focardi, a founding partner of Intertek and a lecturer at the Center for Interdisciplinary Research in Economics and Finance at the University of Genoa, says there has been a fundamental change in the way asset managers regard quantitative management.

"Up to fairly recently the idea of the risk return trade off had not permeated the asset management community. They just were looking for good companies to invest in," he says.

"When they realised they had to consider the risk return trade off, the first thing they did was to place a layer of risk management on top of asset management. So the first use of quantitative methods in asset management was risk control.

"At the same time, at the beginning of the 1990s, a number of people were trying to use quantitative management as a holy grail. They failed, and everybody was very happy to say that quantitative management is only good for controlling risk and for passive management.

"Since then people have come to understand that you can obtain the same creditable results with quantitative management as with active fundamental management."

This has transformed the asset management industry, he suggests. "Asset management has made the transition to an industry with repeatable results and the ability to measure results - everything, in fact, that is typical of an industrial process. This has happened principally because the models used in quantitative techniques work now."

The fact that they work better than they did before is due to the growth in computing power and data, Focardi says. "Seven or eight years ago the type of computing power that is now routinely available to every asset manager was only available to the major players Today the cost of the IT infrastructure is relatively small.

"We also have more data.Data is still expensive but the cost of normal frequency data can now be afforded by even small players."

At the same time, the models available to asset managers are now simpler, and expectations are more realistic, he says.

The Intertek survey shows that regression is the most widely used modelling methodology, followed by momentum. "A number of people are experimenting with more complex models, but everything is now more controlled," says Focardi. "This is part of the general movement towards an industrial view of investment processes.

"Industrial processes are due to good results, and good results are due to the fact that we are now using simpler and more robust techniques and expectations are in line with what models can tell us." Modelling is essential to pension funds in particular and institutional investors in general, if they want to get the trade-off between risk and return right, says Focardi. "The use of modelling is not to produce superior returns. It is not a licence to print money, but it is a way to control investment in terms of both risk and returns.

"Typically, pension fund managers will engage a consultant to create an overall asset liability picture. This ALM picture will then have to be translated and articulated into specific asset management tasks, but with specific and well defined risk return profiles. That is what modelling can now deliver," he says. Pension funds need to take a further step, he says.

"Pension funds typically have a stream of liabilities, so an investment strategy has to be measured not against not the static benchmark but against the liability benchmark. Liability benchmarking is however is a technology that is more complex than the static technology because you have to optimise, taking into account the future liabilities."

 

One of the main findings of the Intertek survey is the use of optimisation. A large majority (90%) of the asset managers surveyed uses some form of optimisation, with most (83%) using mean-variance. This contrasts with the 2003 survey where most firms said they eschewed optimisation.

"A few years ago optimisation was dubbed ‘error maximisation' because of fears that the process would magnify small errors." Focardi says.

"Now that is no longer true. People have understood how to tame optimisation and make robust estimates so they no longer have any estimates or correlations that are totally outside of reality."

For liability benchmarking, asset managers need to take one step further, he says, moving from standard optimisation to stochastic multi-stage optimisation.

"This technology is very difficult for two reasons. First because it's mathematically more difficult than standard optimisation but particularly because, when you look into the future, taking into account the stream of liabilities you have an explosion of the number of facets that you have to consider. So this technology is in a phase of refinement.

"Yet it is a technology that will have to be used if managers want to bring to this type of asset management a similar level of automation that is now used. Looking to the future, liability benchmarking, and with it stochastic optimisation, is the real challenge that lies ahead."

The movement into quantitative management is driven by both the need to reduce costs and the need to achieve better, more consistent returns. Quant managers have shown that they can perform as well as active fundamental managers, Focardi says, and it is the performance of asset managers who use quantitative management techniques that has compelled non-quantitative asset managers to adopt them, Focardi suggests

"The transition from non-quantitative management to quantitative management does not happen because of a progressive transformation. It happens because asset management creates separate groups of quantitative management and they compete with more traditional groups for funds.

"The diffusion of quantitative methodology is essentially due to the fact that in this type of competition the quantitative funds fare quite well. So they have been able to attract more funds.

"The quant industry has evolved because new groups have been created, new people are being injected into old structures and these people are being able to attract the funds to manage."

Eric Sorensen, president and chief executive officer of PanAgora Asset Management, a Boston-based quantitative asset manager, and former director of quantitative research at parent company Putnam Investments, says quant firms are now giving fundamental managers a run for their money in the competition for actively managed funds

"What has happened in the last 15 years is the desire to seek higher returns adjusted for risk, and quantitative strategies have actually moved into those realms. If you look at the US over the last three years, for example, and see who has been hired for active, reasonably good fees to manage money in S&P 500, you'll see they're all quantitative firms." A similar process is under way in Europe.

The reason for this, says Sorensen, is the growing quant literacy in the investment community. "With the broadening of the understanding of risk control techniques, the ability of the pension fund sponsors' consultants who have all of these tools right at their side, and all the money managers, from hedge funds down to indexed enhanced strategies, who have the same toolkit, there's now a great deal of ability to separate skill from the risk that's being taken. That has played well to the capability of the quantitative firms.

"With the tools that are available today a good quantitative form can understand and articulate why they're not performing well and identify the sources of temporary underperformance. They can also explain where they are performing well and why. That's because they designed the portfolio with specific metrics and rankings, so they can be more explicit about the process."

One of the virtues of quantitative management is that everyone can see how it works, Sorensen suggests.

"One of the ironies is that 20 years ago people called quantitative management a black box. But for us it's a glass box. It has logic and it is backed by financial theory, so there's a lot of transparency."

Another following wind for quant techniques is the growing interest in shorting - selling stocks you do not own. The attraction of shorting for portfolio managers is that it enables them to escape the constraints of benchmark weightings.

"Shorting is already part of the quantitative process," says Sorensen. "It's just that historically pension funds have said ‘don't short'. So the quant was always handcuffed. Now they have the machinery available to them to lift those handcuffs."

Quantitative management is also particularly suited to shorting and long-short strategies because it takes an impersonal, uninvolved view of stock selection, he says. "What's so fascinating about the quant process is that it is so dispassionate. It doesn't care. It doesn't say we're not going to short that stock because we know the CEO.

"So we run a global portfolio, constrained long-short, where we are continuously looking at 4,000 highly liquid stock that can be shorted efficiently."

This level of diversification offers investment managers valuable protection when shorts go wrong, he says: "Short positions can hurt you if they're lumpy. If you have just a few short positions, as a fundamental manager might have, you're not going to be right on all those stocks. All of a sudden you can lose all your return on a single bad event. A quant manager can run a portfolio with a lot more short positions and with smaller bets." Quant techniques are also well suited to hedge fund strategies, in particular market neutral, he says. "Market neutral is basically what quants have been doing for a long time."

 

he main challenge for quant managers today is to be able to capture in their models some of the dynamics that fundamental managers consider in their stock-picking.

"When I started doing enhanced models 20 years ago to find a little extra return, people were beginning to look at ways of ranking securities quantitatively and it was almost predominantly value, and maybe some other things like momentum.

"At the end, there was one equation that people would use, or similar equations. There were four or five factors and you would weight these factors and apply that same formula to every stock."

Sorensen calls this the on-size-fits all approach. More recently he and his colleagues have developed a more sophisticated approach known as the ‘contextual'.

As its name implies, this considers the context in which securities are ranked. For example, value, as a selection variable, often depends on the type of firm, the investment horizon, or other non-value factors such as corporate governance and management behaviour.

"This approach is much more dynamic because it actually looks at different pockets of inefficiencies," says Sorensen. "If corporate governance or management behaviour is more critical, you want metrics that apply to that pocket.

"For example, we have a set of metrics which try to assess shareholder-friendly management. We use to call this quality, but we break quality now into balance sheet income statement stability or accounting, the high return on equity. But really it's about management decisions.

Five years ago quant firms didn't have the ability to do that, he says. "We'd look at a cheap company, a company with good quality - that is, a company that had delivered high ROE - and that was it."

"Now we go well beyond that with much deeper assessment. Not as deep as a fundamental manager who would know the names of the children of the chief finance officer or how much inventory they have at a particular plant at a particular point in time. But deeper, much deeper than a simple four or five factor linearly weighted model." Sorensen readily concedes that both quant and fundamental approaches have their place in investment management, and that quant will always be perceived as the duller of the two. He likens strategies such as index enhancement to baseline tennis. "Years back, tennis players were all baseline players with rallies of 20 to 30 volleys to a point. Now you have players who serve and volley. They go for the winners.

"If you own a 25 stock portfolio, you're probably going for the winners, and it's real exciting. If you hit a winner, you can talk about it, and why you bought that stock. But in quant land it's baseline tennis. The aim is to make the opponent make an unforced error. The opponent is the S&P 500, and there are a lot of mistakes in there and we want to avoid those mistakes."