What differentiates asset managers today in the eyes of investors? Convergence is a now a common feature of asset management, with alternative and traditional asset classes now forming part of a diversified investment approach. And as technology developments proceed at a rate of knots, with artificial intelligence and machine learning having become ‘mots du jour’ in recent times, the pressure on managers of all shapes and sizes to deliver smarter performance has risen inexorably.
It was against this backdrop that Global Fund Media hosted a lively breakfast briefing last week, on Thursday, 9 March at the Reform Club in London. The event was hosted in partnership with FIS, a global financial services technology company, with a focus on technology, consulting and outsourcing solutions that serves more than 20,000 clients in over 130 countries worldwide.
Moderated by James Williams (pictured), Managing Editor of GFM, the panel featured: Keith Haydon, CIO of Man FRM & Member of Man Group’s Executive Committee; James Holloway, Founder & CIO, Piquant Technologies, a London-based multi-strategy quant fund; Andrew Davies, Senior Managing Director and Portfolio Manager, CVC Credit Partners, and Ian McDonald, Deputy Head of Aberdeen Solutions, Aberdeen Asset Management.
These are exciting times for fund managers as the early shoots of central bank divergence in monetary policy create more fundamentally-driven markets in which to seek out opportunities – both long and short – in global equity markets.
Over the last couple of years, generating alpha in the current marketplace has been a tough challenge for many asset managers. Central bank intervention has caused increased correlation across assets and made it a real head-scratching exercise as managers try to seek out differentiated returns. Speaking recently to GFM, one New York-based hedge fund manager commented: “The last couple of years no one has had any good ideas. All anybody wants to talk about is whether Janet Yellen’s foot will be on the accelerator or the brake.”
Man FRM’s Haydon made the point that there has been a lot of ‘negative alpha’, mostly based on long/short equity markets “and most funds had the wrong book for that market. However, given that rates have started to rise, this could provide better opportunities to generate alpha.”
Holloway observed that risk has been “spectacularly mispriced” over recent years. This has caused governments to lever up massively.
“Rates will go up and countries will have to pay up; the chickens will come home to roost. Any form of market mispricing eventually leads to a reversal, in this instance this will occur with risk and rates. It’s not a case of if, but when it will happen. It will present a tremendous opportunity – the trade of a lifetime,” suggested Holloway.
With so much political turmoil in the last 12 months, and more expected in 2017 as the European elections roll out, many active managers will hope that volatility returns to the marketplace. If one looks at the CBOE VIX Index – also referred to as the ‘Fear Index’ – it spiked following the Brexit vote last June, reaching 25.76, and spiked again in November to 22.51 following the US election, but in general it has remained range-bound between 10 and 15 for quite some time.
Subdued volatility is, in general terms, the enemy of those wishing to seek out smarter performance. CVC’s Davies is very much a hands-on discretionary portfolio manager. He notes that the European credit markets have actually proven to be useful in terms of seeking out distressed opportunities. “The flow of risk and profiles have shifted. Capital is moving in the right direction (away from the banks and towards institutional investors like CVC Credit Partners) and we believe the fundamentals will continue to be strong in credit risk.
Indeed, the CVC Credit Partners European Opportunities Fund is designed to provide a dual-engine of alpha generation at a time when investors are keen to look beyond traditional fixed income markets. It does this by buying up floating rate senior secured loans across the first lien of the capital structure, targeting a 5 per cent income annual return.
“In addition, through the strategy we are also seeking to generate circa 3 to 5 per cent in capital gains by opportunistically purchasing debt instruments at a discount to its redemption value prior to maturity,” said Davies.
Aberdeen’s McDonald said that they way view the alpha generation environment was from a multi-asset class perspective. He says that a relative approach is often taken, and that despite still not taking a huge amount of risk, the team currently favoured Japan versus the US, “given the recent rises in valuation, a lot of equity market views are tied to the currencies, increased volatility in currency markets and interest rates.”
Keeping a leash on the machine
One of the liveliest parts of last week’s panel debate was the increasing role of machines and artificial intelligence in investment management. Will machines lead to more efficient markets and make it harder for discretionary managers to seek out returns? Will machines fall victim of the crowded trade and lead to more examples of the Flash Crash?
“I’m a strong believer that the trend towards increased systematic trading will continue,” said Haydon. “Computers are getting a lot smarter, people are not. Man Group’s commitment to machine learning (in its Man AHL division) is significant.”
Such is the efficiency and speed of computers and their ability to absorb and process vast amounts of information, the challenge of discretionary fund managers when looking to generate smarter performance is how to ‘outsmart’ the machine; a human might be able to think five moves ahead in a game of chess but a computer has the ability to assess all possible moves playing 100 games of chess.
“Emotion and discipline are not factors as they are for humans,” said McDonald. “A computer doesn’t respond to the latest headline in the same way a human does and is not affected by market sentiment and headlines. The net result, however, is that none of this changes the fundamental laws of economics or finance. The invention of the airplane didn’t make gravity disappear, it just meant we were able to better utilise gravity to get from A to B. Machines are not a magic bullet and AI will not make for better returns alone.”
One of the intriguing aspects of the increasing man/machine dynamic that Haydon touched upon was when to stop the machines. How does one figure out – as an allocator like Man FRM – how to switch them off and control their behaviour for optimum investment gain and minimal loss?
People ultimately will still have to be involved. They write the algorithms in the first place.
Holloway was quick to point out that whereas a machine can be switched off without issue, it is quite a different issue when it comes to the human portfolio manager. “One cannot ‘switch off’ a discretionary manager if they start behaving sub-optimally, that would be cruel,” said Holloway, tongue in cheek.
“However, even if you’re running a fully systemic strategy, don’t be fooled, there is always a human input and a person involved. As with any intelligence, Artificial Intelligence is not able to exist in a vacuum. If a human writes a model that tells a machine how to think, their personality and investment bias is translated into the programming. Therefore, an investor needs to meet the human who is writing and running the programmed strategy.”
Smarter performance might well come from systematic strategies but there will necessarily need to be a teacher/student relationship between man and machine so that the machines learn right from wrong (in trading the markets).
Quantitative managers typically run multiple strategies that the model observes across multiple markets, constantly learning and determining which strategies to act on. But in order to learn, the creator has to imbue the system with values.
“Just as humans learn based on the values of parents, teachers, one needs to set a framework for the system to operate within. It needs to know ‘There be dragons’ in certain market conditions so as not to make a bad decision,” said Holloway.
The inference being that machines are a long way off taking over investment management.
They will be intertwined with humans for many years to come and will necessarily be kept on a leash. What is more likely to happen is that asset managers will increasingly look to adopt aspects of AI or machine learning to improve their investment strategies, whilst maintaining a discretionary overlay.
Managers like CVC Credit Partners remain wholly discretionary in their approach to generating smarter performance because so much of the credit market is done OTC and relies on the innate skill of the manager to source the right deals; be they primary or secondary. It is the preserve of the human mind.
“Our system is manually drawn from an in-depth look at every asset that the market is pricing. It has been created over ten years by drawing upon data from fundamental performance. We have 35 analysts doing all the heavy lifting, constantly monitoring fundamental performance. It is very much a human process not a machine process,” explained Davies.
That will reassure some investors who may wonder about the efficacy of machines to truly generate smarter performance in a wildly complex marketplace where, by and large, asset prices are influenced by human emotion.
Haydon added that humans will continue to intervene in computer models because they believe they can do better. “Often though, humans intervene and make things worse. People want to retain the responsibility. We don’t accept computers as a repository of moral responsibility. As we continue to operate a culture of allocating capital, if the human were to be removed completely we wouldn’t trust things,” opined Haydon.
One area of general consensus among the panelists was with respect to ‘smart beta’ as an investment concept. The view was that there isn’t really anything smart about harvesting beta, even if it is done in a more considered way, using various factors such as value, momentum and low volatility to reconfigure indices and build portfolios.
In a Financial Times article (‘Smart beta not quite as clever as marketed’) last March, the point was made that focusing on a single factor is very risky. The article referenced Andrew Lo from the Massachusetts Institute of Technology, who noted that the problem with “smart beta” is that it can easily morph into “dumb sigma”. Sigma is the letter used to denoted volatility.
Asked whether smart beta is oxymoronic, Holloway responded: “Dropping the first three letters would be more accurate. There have been no crises for CTAs but I think it is inevitable that at some point something will go wrong with smart beta. These are passively managed (by and large) and will probably exhibit the same problems as an equity crash; it just hasn’t happened yet.”
Alternative data sets
One important tool that has recently emerged for investment managers is the use of alternative data sets; GPS data, social media data etc., or as Holloway commented, weather data, which can prove highly effective in building investment theses.
“You need to have data that others don’t have,” said Haydon. “But bad data is worse than no data at all.”
Holloway said that data was the “lifeblood of what we do, so we don’t want polluted datasets”.
Looking ahead, McDonald said that regardless of whether the strategy is 100% systematic or 100% discretionary, when it comes to generating smarter performance it all boils down to the initial decisions that are made on how to allocate capital:
“I do not think we will get to a place where computers take over the economy. Portfolio managers will still run investments.”