Quant: Down but not out
Each crisis delivers useful lessons to the quants world, writes Iain Morse. What did the last one teach us about the optimal conditions and most dangerous risks for model-driven strategies?
Something happened during the week of 6 August 2007. A set of what had been the most successful quantitative investment strategies lost money when they should not have done. Weeks earlier, in June, two Bear Sterns credit strategy funds had failed - in July, Sowood Capital Management lost 50% in value; meanwhile the US mortgage market failed. But these were all credit events. In that week in August, matters took a turn unpredicted by quant models.
The worst losses - ranging from -5% to -30% - were in funds employing long/short equity market-neutral strategies, also known as ‘statistical arbitrage’ strategies, which were designed to be relatively unaffected by market volatility because they are not about capturing market beta. These losses were dramatic and humiliating for quant managers used to denouncing their less intellectually rigorous rivals. “High correlation makes markets dangerous for quants,” explains Jean-Francois Schmitt, CEO for UK quantitative strategies at HSBC Global Asset Management.
Analysis of August 2007 has been extensive. A look at the most widely accepted conclusions should tell us something about the mix of quant strategies and risk control techniques now en vogue. But one word of warning: like generals, risk managers have a tendency to fight the last war, not the next one.
First, though, let’s recall the fate of Long Term Capital Management a decade earlier. LTCM’s demise was a result of ‘convergence’, or mean-reversion, trades, principally on bonds. The fund had been ambushed by unique events which, by definition, could not be factored into its models: sovereign defaults, notably by Russia, caused market panic and LTCM huge losses. Quantitative managers learned a variety of lessons and applied them to their risk management processes.
So which event had not been insured against in August 2007? The first key lesson was that losses were initially the result of a group of funds with similar strategies trying to unwind losing positions more or less simultaneously and in volume.
“For certain types of strategy stop-loss is the worst remedy to losses,” warns Jonathan Xiong, managing director of the global asset allocation team at Mellon Capital Management. “In some conditions it may make sense for momentum strategies but not many others.”
Another lesson was the danger inherent in the huge expansion in leverage used by quants. Long/short equity had attracted a huge amount of money over the previous decade. Market-neutral strategies had, for some time, seen their high excess returns eroding. More managers, using similar or virtually identical strategies, were trading a market with declining volatility. Leverage was the quants’ solution for squeezing excess returns from thinner and thinner deal margins. They also held wide, non-concentrated portfolios, the collective disposal of which was always likely to have an equally wide market impact.
“Did anyone realise how leveraged the market had become?” wonders Andy Barber, senior investment consultant at Mercer. “The answer is probably not, as it built up very rapidly.” De-leveraging caused more de-leveraging as the exit was blocked by a fleeing crowd. “They all de-risked the same factors at the same time,” says Xiong. This panic lasted only a couple of days, but although markets recovered most of their losses, positions had already been liquidated and could not be re-established without further substantial loss.
So the key lessons drawn by the quant community this time acknowledge the interrelatedness of financial markets and the dangers of excessive leverage. Both amount to a paradigm change for this group of investors. And the entire episode has a sparked yet another debate over whether quant has had its day.
“The short answer to this is ‘no’ - and we’ve been here before,” cautions Mike Arone, global head of product engineering at State Street Global Advisers. “I have copies of articles dating from the bursting of the tech bubble that covered just this ground.”
Perhaps a more significant debate is that around the optimal conditions under which to run quantitative strategies. Some suggest that moderate volatility is optimal without major inflections or sudden reversals, but not everyone agrees.
“Too much attention is given to volatility, not enough to cyclicality,” Arone insists. “There are some stages or periods in the cycle when quantitative works well and some where it works less well.” But, with so many strategies within the quantitative universe, generalisations are hard to make. Within a diverse universe of 20 or more recognised quant strategies, some put an emphasis on momentum, some on value, some are of very short duration, many are of medium to long-term duration.
“You need to take a nuanced view of this issue,” judges Michael Fraikin, head of client portfolio management at Invesco Asset Management. “The worst environment is one of decreasing volatility. Sustained strong trends are good, and high volatility is good as long as it is sustained.”
There is also debate over the consequences of consolidation in the asset management industry and the rise of low-cost passive management for active quants. The argument here is that the construction of the indices, particularly market cap-weighted ones used as benchmarks by passive managers, can be exploited by quants. “If the move to low-cost passive management continues, it could create more opportunities for quantitative strategies,” Fraikin suggests. “Passive money is dumb money.”
This may be the reason why the shift to passive management has been accompanied by a corresponding increase in allocations to alternative investments. “This is creating a barbell effect, with a squeezed middle comprising benchmark-relative, risk controlled strategies,” Arone argues. “We will make investment decisions on data ignored by the passive managers.”
Indeed, another trend is the number of alternatively-weighted ‘passive’ benchmarks that have been launched which look like attempts to capture the alpha generation of some value and momentum-orientated quant strategies.
A good example is the Research Affiliates Fundamental index (RAFI) series launched by Research Affiliates and FTSE. Rather than market-cap weighting, RAFI have created indices that, like many active quant strategies, select and rank constituent shares by fundamental measures of company value, such as sales and dividends, that the market will ultimately recognise that it is mis-pricing. RAFI claim they can outperform conventional cap-weighted indices by 2-4% per year, a similar range of outperformance as that set for many long/short quantitative strategies.
“They say there are index providers,’ notes Fraikin. “But, really, their approach is a rather crude application of quantitative strategies. I take what they do as a backhanded compliment.”