Quant models limber up for change

Active quant strategies came in for criticism after the global financial crisis, with a number of models seen as lacking both the appropriate diversification and the dynamism necessary to react to major market events.

While acknowledging the need to rethink quant models, global head of active equities for developed markets at State Street Global Advisor (SSgA), Marc Reinganum, tells how quant investing has evolved and how the manager has overhauled its own approach.


Dynamic models

Reinganum says SSgA took a close look at its quant models in the wake of the financial crisis and what emerged was a consensus that, whatever approach was taken, it needed to be more dynamic and able to quickly rebalance to take into account of changes in market conditions.

“Active quant equity at SSgA in 2012 has evolved substantially from where it was in 2007,” he says.

“We have done an enormous amount of research to make our models more agile and nimble, and able to be appropriate for the prevailing macroeconomic environment and investment conditions.”

Reignanum calls these “dynamic models” as they are designed to change as the effectiveness of the characteristics or factors the models use to forecast future performance also varies at different times of the economic cycle.

“We have extensively studied how to systematically relate conditions we observe in markets today to the intermediate performance of these factors,” he says.

“This is typically over a three-to-six-month investment horizon. Back in 2007, we were very focused on long-run models, which in the long run do work. But it became very clear to us that these models built on long-run performance – that is, how our factors perform on average over time – [and] the context of that approach was important.”

These conventional long-run models perform well when market conditions and economic climates are near long-term historical norms, according to Reignanum, but underperform when markets are more turbulent and abnormal.

It was these conditions that prevailed during the first heyday of quant investing, with quant strategies performing strongly in an economic environment that would later become known as The Great Moderation.


Historical data plus…

However, Reignanum says SSgA realised that its quant strategies had to also perform during times of abnormal conditions, when there was greater market volatility.

What emerged was a two-year project to mine historical data from a number of countries.

The 40-person research team looked at how more than 100 factors, such as the momentum and valuation factors of particular stocks, performed over time in an effort to see how that performance could be forecast.

SSgA also expanded the range of factors that it studies.

“Most quantitatively orientated investors, including SSgA, focused primarily only on those factors that worked well on average over time. But as we did our dynamic modeling – and if you think about wanting to understand how models work over intermediate horizons – you need to open up the potential set of factors that you study,” he says.

“This will include factors that may not work well in the long run, but may work quite powerfully over shorter, intermediate time horizons.”

Reignanum says an example of this is financial leverage, where a long-run model would not include financial leverage, despite there being different market episodes where it might be either advantageous to take on leverage or better to avoid it as much as possible.

“In 2012, we allow our models to include these transient or opportunistic factors that were not in long-run models,” he says.

While the historical performance of these factors is an important ingredient in the quant model, it also must be forward-looking and reactive to changes in market conditions.


Beyond the conventional

As it did for investors around the world, the global financial crisis sharply brought into focus for SSgA the need to better understand the broader macro environment and the changes within it.

However, SSgA wanted to maintain its systematic and disciplined investment approach, Reignanum says, and this pushed the manager to look beyond conventional macroeconomic forecasting measures such as GDP, employment or inflation.

“We are looking at how investors actually express their views in markets today and that defines the macroeconomic environment and market conditions – very much market-based and very much not subjective but objective,” he says.

To form this “objective” view of markets, SSgA monitors what is happening in fixed-income markets, examining information on term structure, at credit spreads.

The historical performance of the 100 factors SSgA has identified and the monthly analysis of market conditions are then plugged into a range of forecasting equations that are designed to forecast the future performance of each of these factors based on the conditions that are observed at a given point of time.

The forecasting horizon is three to six months, with forecasts updated on a monthly basis.

“We view this process as expressing high conviction, which is perhaps unusual to hear from a quantitative-oriented firm,” he says.

“The term conviction or concentration has been usurped by fundamental analysts. If you talk to a fundamental portfolio manager, he might often say ‘my best ideas are expressed in these 20, 30 or 40 stocks’ – a very concentrated sense of the world in terms of individual stock holdings. From our quantitative perspective, we view concentration in a different way. We see it as focusing our portfolio on themes or characteristics that are likely to pay off over the next three to six months.”

This involves choosing securities that provide an exposure to the most impactful factors at a particular point, whether that is value, momentum, reversal, risk or risk aversion over the short term.

“Our best ideas are created by using securities to form portfolios that have the desired exposures to desired themes or characteristics that we think will be rewarded. We do not explicitly look at concentration as a limited number of securities, rather we try to get those exposures using as many different securities as we can so we don’t have investor portfolios that are sensitive to firm-specific or idiosyncratic risk.”

The longest model that has been running in using this framework has been a US small-cap strategy.

It began in 2011 and has been a top-quartile-performing strategy out of a universe of about 700 offerings, says Reinganum, performing particularly strongly in last year’s choppy market conditions.

He says the model also predicted that despite widespread uncertainty and market volatility during the third quarter of last year, that markets would rebound and risk would be rewarded.

“At the time the human emotion would say ‘I want to run away from this risk’ but by keeping our rudder deep in the water, by the fact that we knew we had done extensive and thorough research, we knew what direction to set sail and we did. That is the advantage of the quantitative process. We can mitigate the impact of emotions,” he says.

Earlier Quant models, pictured above right, limber up for their creator, Mary.