Sections

Strategically Speaking: Research Affiliates

Have we entered a new golden era for quantitative investing? Trends such as the decline of traditional active management and the rise of machine learning-based investing would suggest that is the case. But Rob Arnott, founder and chairman of Research Affiliates, a self-described “lifelong quant”, seems prudent about the prospects for quant strategies. He finds the quantitative-investing industry often guilty of “overhyping and overselling” ideas. 

Yet the past decade has seen the remarkable growth of quantitative investing. Arnott’s firm, Research Affiliates, has amassed $170bn (€148bn) of assets under management since its foundation in 2002. 

Arnott founded Research Affiliates after a career spent between academic studies and executive positions at various asset management firms. He is a staunch advocate of the use of the scientific method in finance, and has no doubt in the power of asset management to solve financial problems. 

However, he argues that the flaws of quantitative strategies tend to be overlooked. “There is some real naivety in the factor-investing community. It is very sad that – while the ideas, in many cases, are good – they are just being hyped as being more remarkable than they truly are,” he says.

Perhaps because of its roots in academia, quant investing is an industry where ideas are deeply held. Arnott suggests quants may have an ego problem. He says: “I think a certain dose of humility is necessary for successful asset management. One must recognise that any strategy will be less than perfect and it is likely to have an array of errors or flaws that might have been avoided.” 

The industry is having one of its biggest debates precisely around its relationship with academic research. Purists advocate the use of strictly academic definitions of factors, while most practitioners use more complex definitions. The debate matters because different definitions of factors in portfolios can provide different investment results. 

Having authored more than 100 papers in academic journals, Arnott is sanguine about his firm’s relationship with academia. He says: “Those who say that you have to hew to the academic definition of a factor, in order to deliver the same results as the literature would suggest, are overlooking some very basic problems.”

screen shot 2019 04 30 at 15.00.01

First, the fact that the academic literature shows that a factor provided a particular return in the past says little about how well the factor will work in the future. “I am constantly astonished at how many quants buy into the notion that a backtest is a perfectly good way of gauging future returns. I have never seen so many billions of dollars invested into strategies based on a backtest,” says Arnott.

“Secondly, the academic papers are written by academics, many of whom have never traded a stock or been involved in institutional-scale asset management,” he says. This means they have little understanding of the impact of trading costs on factor strategies. 

“The third problem is the tacit implication that the research has no merit when it is not done by an academic, even if a practitioner designs a factor that is more powerful and more liquid than the academic factor.”

The vast reliance on backtests is the biggest problem. “If a stock has been massively profitable in the past, chances are it is because valuation levels have soared, and that might be a sell, not a buy signal. The same goes for a factor,” Arnott says.

He does not hold back in his critique of the industry. “Scientific method is lacking in many areas,” he says. “Watch how many factor strategies that work beautifully on paper have been launched, only to find that they do not work on live assets. Instead of asking what part of the logic was wrong, the question that is often asked is, how do we improve our model so that it would have worked recently, too?”

The value factor, which purists define as the ratio of book value and stock price, is a good example of the naivety of the factor-investing community. “Taking our company as an example, our most valuable resources are our staff and our most valuable product is intellectual property. Our company’s value is not remotely captured by book value.”

Research Affiliates’ definition of value, which is complex, appears to work remarkably well. Arnott notes that the firm’s RAFI Fundamental index is among the tiny minority of value-based strategies that have outperformed over the past decade, a famously disappointing one for the value factor. The index has outperformed emerging market indices by 130bps, has added 70bps to developed market indices (ex US) and 20bps to US indices, according to Arnott.

Contrary to many factor investing strategies, the starting point of the strategy is not a market capitalisation-weighted index. Arnott argues that this makes it a smart beta strategy in the proper sense. “The original definition of smart beta was a strategy that breaks the link with price and therefore has a rebalancing alpha,” he says. “Today, smart beta is often used as a catch phrase, but it used to have a tight definition when Towers Watson created the expression in 2007.” 

This rigorous but pragmatic approach to research has delivered good results for investors in Research Affiliates strategies. This means the company is under less pressure to rush to adopt machine learning. Arnott regards the approach as a “very powerful set of tools that can be used to better understand how market functions”. However, he is sceptical about its application to monthly and quarterly data that is required to build factor-investing portfolios.

When it comes to machine learning, and asset management more in general, Arnott says: “the goal is to have a checklist of ways you can hurt yourself. When dealing with big data and machine learning, that puts modelling errors on steroids”. A framework for building that checklist was published last year in a paper that Arnott authored with Nobel Laureate Harry Markowitz and Canadian economist Campbell Harvey.  

The asset management industry is facing unprecedented challenges. Arnott says that Research Affiliates is equipped to meet them. “We are in an odd position of having very nearly the highest assets under management we’ve ever had, with revenues that are down from five years ago,” he says. This is thanks to the “price war” that is happening in the industry.

“We prize our independence and our corporate culture, which rewards quality research, over and above short-term business development and growth. As such, our focus is aligned with our clients. We care deeply about how well our strategies will work, because that, ultimately, is the engine for long-term growth,” he says.

Have your say

You must sign in to make a comment

IPE QUEST

Your first step in manager selection...

IPE Quest is a manager search facility that connects institutional investors and asset managers.

  • QN-2563

    Asset class: Mid & Small Cap Equities.
    Asset region: Global.
    Size: USD $130m.
    Closing date: 2019-09-27.

  • QN-2564

    Asset class: Large Cap Growth Equities.
    Asset region: Global Developed Markets.
    Size: USD $130m.
    Closing date: 2019-10-04.

Begin Your Search Here
<