Man and machine working together

10 August 2009

Financial markets have evolved to allow electronic trading, which excludes the possibility of human error. But machines should not replace managers; they can work alongside each other.

Quantitative asset management is by no means a new concept. It is rooted in the work of Harry Markowitz, a Nobel Prize winner, who first proposed modern portfolio theory nearly five decades ago. In America, quantitative asset management enjoyed rapid growth, twice as fast as that of classic asset management over the past three years, and is now estimated to account for 20% of equity funds. Quantitative teams are being established at an increasing number of institutions.

Financial markets have also evolved dramatically over the past few years to allow electronic trading on most markets, to handle short positions and to access an array of new asset classes. The evolution of Ucits III regulations has also been beneficial in broadening the eligible assets, for example, the integration of listed and over-the-counter derivatives.

In simple terms, quantitative asset management uses algorithms based on a set of formulae to invest, leveraging on the calculating power of computers. One of the big differences when comparing this approach with traditional active management is that it removes discretion, and it can answer certain needs that discretionary management cannot.

Investment decisions are fully process-driven, governed by an algorithm and applied on a systematic basis. This allows dynamic portfolio adjustments without discretionary intervention and therefore excludes the likelihood of human error.

With quantitative models, investors are clearly not buying something as complex as the human brain, but this brings some advantages. The idea is not to replace humans with machines, but to remove the emotional aspect and behavioural pressures that all asset managers face. There is, in theory, more stability. Quant models use objective criteria only, so they do not have to react to markets and investors know exactly what they are investing in. By using numbers, the process becomes very defined – the numbers do not leave any space for interpretation.

There has been some concern over the complexity of algorithmic investment, but in practice it is actually the opposite of a black box. As the algorithm is known beforehand, and both the models and the rulebook can be disclosed, systematic strategies offer high transparency. There are no surprises as investors have a clear view of positions and potential risks at all times. And with open disclosure of management models, it is easy to read performance in any given market condition. Models have also enabled portability, bringing the ability to spread expertise from one market to another.

In the aftermath of the recent financial crisis there is a strong need for risk control, in addition to liquidity and transparency. Many investors who previously used hedge funds, for example, are using quantitative models. Where previously it was all about performance, they now want to know what is behind the black box of the investment process.

Risk controls, or volatility control mechanisms, can be integrated within the process – an important feature particularly in today’s challenging and highly volatile markets. Not every strategy comprises a risk control mechanism. However, thanks to technological advances and stress test scenarios, implemented risk control measures are much more effective than an active manager.

Furthermore, with the help of computers, new systematic strategies have been able to optimise risk/return profiles with more sophisticated diversification and de-correlation methodologies.

Investors are increasingly keen on quant-driven multi-asset allocation strategies. With the help of technological advances, a quantitative model can easily determine the optimal weights of a defined set of asset classes to maximise the expected performance with a given level of risk. Portfolios can be dynamically rebalanced on a frequent basis. In addition, risk-adjusting measures (built-in volatility control mechanisms) can offer protection to adverse market conditions.

The algorithms used in quant-driven models are based on long-term, proven methods that take into account the work of people who have been in the field for many years but without the discretionary intervention on a day-to-day basis.

Another strategy which has proved successful, particularly in bearish and stable markets, is covered call writing, also known as buy-write strategies. This involves the combination of a long exposure on an underlying and at the same time selling call options on the same underlying asset. While the concept is nothing new, it can be improved via quantitative asset management. The frequency of the overwriting can be increased, up to a daily basis. This improves stability while protecting against strong market rebounds.

Commodities are another asset class of choice for systematic strategies. Quant-driven strategies can help optimise the roll returns for each commodity and dynamically adapt to changing market conditions, as commodities tend to exhibit market trends significantly.

Roll returns involve commodities bought via contracts maturing at a certain date. The exchange of an expiring contract for a contract expiring at a later date is known as the roll process. These roll processes vary for each commodity. As prices of commodities can fluctuate from one contract to another, certain risks are associated to commodity-based strategies.

Traditional asset managers undoubtedly will always have a place, but they do come with some manager-specific risks.

Investors trust a manager based on his track record and reputation, but there is always an inherent risk that the strategy could drift from the investment mandate – something which is removed completely in the quant approach. In a sense, quantitative asset management is complementary to active or passive asset management.

While quantitative models should evolve regularly, it is important not to adjust them too often as this might introduce discretionary intervention. The key is to adapt models to a changing world. To exploit quantitative models to their full potential, providing investors with maximised returns, efficient risk control, liquidity and transparency, it is essential to have a high level of technical expertise and experience.