The Case for Re-Evaluating Quant

What every investor should know

Originally published in the May 2008 issue

The headlines could hardly have been worse in August of last year and since then, some of the mainstream press, armed with a little knowledge, (always dangerous!) have sought to pin almost any market disruption on “those secretive quant managers with their black boxes.”

Of course in reality, as most hedge fund investors know, quantitative managers operate highly disciplined and diversified investment processes. However, there is a vast range of strategies ‘bucketed’ into the Quant category with some popular misconceptions applied across the board. Essentially, there are two main equity quant hedge fund approaches; statistical arbitrage – trading strategies that started life in the investment banks’ proprietary trading divisions and fundamental factor-based strategies – more typically employed by the large asset managers. These two strategies are not mutually exclusive and, in fact, if both are combined to an optimal level, a consistent return series is exhibited with possibly less expected contribution in extreme market volatility spikes but, achieving a far greater return series over a cycle as many more opportunities for profit in most market environments are generated.

Statistical arbitrage means reverting strategies tend to be greater return contributors in times of higher volatility and can therefore provide portfolio insurance. However, they can underperform when factors are trending. Traditional static factor strategies tend to perform strongly in rational markets and particularly when investors are focusing on value and momentum-based themes. However, these strategies struggle to achieve returns when crowding occurs.

A 21st Century generation of factor strategies, however, have additional degrees of freedom and can capture returns from being both long and short factor exposures over a variety of different time horizons. These dynamic strategies are paid by collecting the premia due to the longer term economic cycling effects and the premia due to shorter term behavioural effects. Incorporating contrarian strategies into the optimised selection of dynamic factor bets enhances the overall robustness of the strategy and provides downside protection at inflection points. Thus, these strategies aim to be ‘all weather quant funds’. So quantitative strategies are clearly not all the same. Some are designed with higher risk appetites in mind and others are constructed to appeal to those with a more institutional investment requirement ie. consistent returns with low annualised volatility. Some strategies are highly complex, some are reasonably transparent and some have elements of both attributes. Keeping it simple, as with any process, means that there is less to go wrong.

Mathematically designing a straightforward process to generate the best possible returns and being able to communicate that strategy to investors provides comfort as one is able to ‘see inside the box’. The further enrichment of the core strategy by incorporating additional intuitive models, developed from proprietary data capture, provides investors with the extra assurance of knowing that their chosen manager is not necessarily ploughing the same field as others.

Essentially, however, quantitative strategies do have one common aim; to maximise the accuracy of forecasting skill – getting a greater number of bets right, and in greater magnitude than those that are wrong.

During the last few years there has been significant growth in the size of assets managed quantitatively in long only products, ‘hybrid’ or 13030 strategies and, of course, hedge funds. Many argue that this is what led to the contagion effectin August of 2007 – so many managers pursuing value and momentum-based methodologies.

So what should investors be looking for in quant strategies going forward?

  • A long track record is a good start – preferably 5 years or more. Managers that can demonstrate consistent performance over a variety of market cycles – including bull, bear and ranging markets will provide investors with a good framework of what to expect and how to incorporate these managers into their portfolios.
  • And concomitant to the above, an experienced investment team with many years history of working together – with low turnover of the core team being essential. Experienced managers largely survived the August turmoil as they were quick to assess that their models were not broken and that the ‘disruption’ was a forced liquidation event that would be short term in nature.
  • A bespoke risk management structure. Anecdotal evidence suggests that possibly one of the reasons for the ‘August Effect’ was that several large managers employed standard factor risk indicators. Risk management should be integral to the alpha generation process – liquidity criteria, factor, sector and stock exposure constraints etc driving portfolio allocation.
  • Unique factors models. As more money is put to work in factor strategies, managers focusing their research and development resource on identifying and building ‘proprietary’ factors will have an information advantage over their peers. Innovation instead of imitation is vital.
  • Quantitative global portfolio allocation. A discretionary regional allocation is sub-optimal as it can lead to trading too great a share of the portfolio in unprofitable markets for an over long duration.
    * Accurate estimation of capacity. This is essential – if the alpha decays due to a weight of money in the strategy then greater leverage will be employed to meet investor’s expectations. A nimble fund will meet its stated risk target using moderate leverage and will be less vulnerable in a liquidity trap.
  • The strength of the process. Over-fitting/data mining is a common problem in back testing quantitative strategies. Models should not be decaying on a frequent basis if the strategy is robust. Key in forecasting is solving for the most suitable time frame for determination of alpha values. Achieving the balance of avoiding using a time series that places overemphasis on fleeting periods of over performance but, not using too coarse grained an approach so that a rising star is missed until it has achieved all of its gains, is critical to successful quant strategies. Daily and monthly data series both have merits and drawbacks. Optimal is most likely a twin track approach.
  • Dynamic strategies. Adaptive factor strategies can react quickly to changing market sentiment and capture the new winning themes. Key is the ability of the ‘regime change’ models to be sensitive enough to capture real signals but not so sensitive as to react to noise. Most quantitative funds do not use dynamic factor weighting schemes due to the difficulty in developing back tests where the dynamic scheme works better than a static one. However, as was evidenced last year, utilising dynamic weighting schemes opens up the opportunity to use a far broader array of factors and to take long or short exposure in some cases. It can also be seen that, in dynamic strategies, certain factors provide a natural hedge for others in the portfolio, thereby reducing risk in extreme events.

About Sabre Fund Management

Sabre’s Style Arbitrage strategy was the first dynamic factor strategy in Europe when it was introduced in mid 2002. Sabre’s dynamic factor strategies have generated 14.18% and 18.73% in 2008. Since inception, Sabre Style Arbitrage Fund Ltd has achieved its stated compound annualised return target of 6-8% over cash with a Sortino Ratio of 2.7 (for period from inception in Aug 02 to Mar 08.