Cover story: Applying science to fund management


Today, we live in a world where the overlap between man and machine is getting bigger and much theory is being applied to how data and decision are not mutually exclusive.

Technology is improving on a daily basis, and costs are coming down, which, when coupled with the propulsion of the internet, means the amount of data is growing exponentially. But what can we do with it? Are we at risk of data saturation?

David Wright, EMEA head of product strategy for BlackRock’s Scientific Active Equity team, believes there are twin benefits. He says: “It allows you to access not only a [greater] level of information, but also new types of information that were previously unavailable.”

The SAE team is responsible for $85bn in assets under management, held across a suite of quantitative-based funds.

Wright says the simple beauty of quantitative, or systematic, approaches lies in their potential reach.

“It’s largely about breadth,” he says. “A traditional manager might be able to cover hundreds of stocks, but we can cover thousands – albeit perhaps with a ‘lighter touch’.”

Richard Taffler, professor of finance at Warwick Business School, recognises big data as one of the more trendy – while superficially plausible – themes in investment circles today. An international authority in behavioural and emotional finance, he thinks such stories help make people feel better about participating in something that is effectively based on uncertainty, and help us make sense of the world.

He says: “You want to believe that by using the power of computers, enormous amounts of social media data and the magic of data scientists, you can beat the market – despite there being no evidence for this. It’s all very fashionable and exciting, with the information provided by [social media] but can it really help you make better decisions that lead to superior returns?”

Such techniques may lead to outperformance but he believes it is likely to be by chance “because distinguishing between skill and luck is very difficult in practice, if not impossible”.

John Husselbee, head of multi-asset at Liontrust, sees a growing use of quantitative approaches, but primarily for reasons of efficiency – with a trend towards a “halfway house” emerging.

He appreciates that if embracing technology allows a fund manager to increase productivity, it means time and efforts are far better spent. “If you can use algorithms to filter down your ideas, it gives more time to apply your process and judgement to a smaller number of stocks, rather than wasting time on ideas that might have been dismissed sooner by using a quantitative approach.”

For a number of years, as technology has become more widely accessible at a more reasonable price, he has seen a shift taking place from a purely fundamental approach toward quant-based models as a filter if nothing else.

There is a lot of scope to find a natural meeting point between man and machine in financial services, from the rise of automation and robo-advice or the introduction of artificial intelligence (AI) and wearable technology tracking activity.

Clare Flynn Levy is founder and chief executive of Essentia Analytics. The firm has brought together experience in fund management, technology and behavioural science to produce software designed to identify patterns of investment behaviour, which are then analysed to help improve future decision-making.

She believes the hype around AI is premature, and the notion of automation can confuse as to where each stakeholder – end client, adviser and fund manager –can truly add value. Where there is valid scope, she says, is to let
humans carry out the tasks to which they are better suited – the actual decision-making, and take away from them those aspects where a computer may do a better job, such as number crunching or detecting anomalies. Recognising the need for a consistent, repeatable process, fund managers are often on the lookout for ways to “dehumanise” their approach.

Nick Kirrage, fund manager in the equity value team at Schroders, places a lot of emphasis here. He understands that as a value manager, and with the proliferation of smart beta-type products, there is an increasing requirement to justify the value they add as human, active, fund managers.

A screen can produce a list of the cheapest stocks. For instance, in 2009 it would be a list full of financial names. As very few clients would be comfortable investing in such a list, one of the fund manager’s jobs is to diversify that exposure.

“The other is that we’re looking to make better risk-adjusted returns,” he adds. “Sometimes you‘ll look at the screen and say, ‘I genuinely think investing in HMV is a bad idea because I don’t see the future of CD retailing’.”

Further, he says quantitative screens may not be able to distinguish between two equally priced companies.

“Depending on which screen you’re looking at there’s quite often no differentiation between the balance sheet risk you’re taking. It might just say ‘this is cheap’. Two companies could have equal ‘cheapness’ but one could be in a dire financial situation and the other have a reasonable financial situation. Of those two I know which I would rather buy or which my clients would rather I buy.”

Wright agrees: “It’s not about just using the machinery and applying it to the data. There is a human element. It’s the blend of technology and investment insight, not just pointing machinery at data sets.”

IBM estimates that we humans generate 2.5 quintillion (or 2,500,000,000,000,000,000) bytes every single day. And 90 per cent of the data that exists in the world today has been created in the past two years alone. Yet what is the point in having so much information unless it can be organised and used effectively?

Husselbee says: “In our industry we seem to have lots of data but not much that is particularly useful because it’s all a bit one-dimensional. We measure a fund manager’s level of success not how that success – or failure – has been achieved.”

In sport, coaches often play back footage of previous games to their teams to identify what may have gone right or wrong, boxers repeatedly watch video footage of their fights to spot their vulnerability or repeated mistakes, and in golf, coaching techniques rely heavily on footage of past performance.

Why should fund management be any different? It shouldn’t, according to Flynn Levy, who says: “The use of data in optimising pretty much any skill-based activity is a trend that you can see across any industry. The more data that can be collected, the better the feedback that can be delivered in order to make things better in future. And that goes for anything from a car assembly plant, to a professional football coach, to a professional investor.”

As the amount of available data grows, as does demand for managers to demonstrate a robust and consistently applied process, gaining an understanding of what works and does not work will go some way toward determining whether a manager is performing due to luck or skill.

Kirrage uses data selectively. While not necessarily seeing huge benefits in scraping airline ticket data for pricing trends, for example, he does see the value in gathering data that surrounds his own decision-making.

“How do I analyse every decision I’ve made over the past 10 years, work out which were the better decisions and see if there’s a relationship? What were my worst decisions? And were they down to my having too much coffee that morning, or not enough sleep? Some of those things are very hard to track and what I would say is that it is very easy in fund management to confuse causation with correlation.”

While many active fund managers sing a mantra of being skilled in idea generation, Kirrage says his (and co-manager Kevin Murphy’s) process is its antithesis. Supposing you are “skilled” at coming up with ideas suggests they are choices, or judgements. Whereas a way to avoid such psychological traps is to use screens, which he says strip out human emotion.

“Then there is no liking or disliking,” he says. “Our idea generation literally starts with a list of companies that will be the cheapest stocks in the market.”

His job is to determine why those stocks are cheap, and figure out which ones have an investment case worth considering. “We’re not cyborgs; psychologically, we are who we are, but we’re just very introspective about why we feel happy or sad about stuff and are constantly trying to remove as much of that as possible from the way we invest other people’s money for them.”

So in applying a data-driven process and formula to psychological behaviours, does that make them less prone? Behavioural finance suggests we are hard-wired toward biases – they are instinctive. Yet in understanding them, are we truly able to overcome them?

Professor Taffler believes we make all decisions unconsciously and that it takes seven seconds for those instincts to reach our conscious mind.

“Fund managers want to believe they can be rational, and often make many attempts to ‘deny their humanness’”, he says. “But they can’t – they are human and therefore predisposed toward emotional decision-making.”

Some level of crossover seems to be the sweet spot. According to Flynn Levy, frustrations exist at pure systematic houses due to the intrusion humans can bring to their black box methods.

“No one is spending time to measure whether the human intervention is even adding any value,” she says.

But as humans are the ones responsible for writing the algorithms and putting together the strategies in the first place, is it even possible to entirely strip out human bias?

“How do you know there is no bias in your algorithm? You can only know that if you go through the same sort of process, logging decisions and identifying patterns of behaviour,” she says.

In an increasingly competitive space, especially as the active versus passive debate continues to rage, any edge an active fund manager is able to gain is surely worth the effort. What companies like Essentia and Inalytics have identified is that the more data one can gather about the decisions a fund manager makes, the greater the likelihood of them spotting any habits – good or bad.

Common habits seem to relate to timing. For instance, as per the common gambling analogy, when an investor has been on a winning streak, they tend to gain confidence and overextend – in behavioural finance, this is known as the “victory effect”.

Flynn Levy says: “The trick is not to try and remove inherent biases – we can’t ask you to not feel them, that is virtually impossible. The trick is to try and stop acting on them. We believe the best way to identify those is to hold the mirror up. And then work out how much those habits are costing you.”

Once certain problematic behaviours have been identified, they can be caught with nudge messages and prompts sent out at times when things might – according to one’s trading history – become a bit tricky, or vulnerable.

As investment markets can be so fickle, Professor Taffler believes fund managers can face great difficulty in beating the market on any kind of consistent basis. Further, the anxiety caused by such challenging conditions is often unacknowledged.

“Future investment returns are uncertain and unpredictable, yet investment management is sold on the basis that fund managers are able to do what is not possible – that is outperform other managers or the market on a consistent basis. This is mostly because they are very good at buying stocks but are generally very bad sellers. So if you can learn to be a better seller, your overall fund performance should improve.”

It is precisely this relationship between data and psychological bias that Inalytics is trying to harness. Chief executive Rick di Mascio says it plays firmly to the “disposition effect” – essentially where people want to protect profits and deny losses.

“Fund managers tend to sell their winners way too early. Instead of doing what they should do, which is to run their winners and cut their losers, they very often do the opposite, which is cut their winners too early and hang onto the losers too long. I see data and psychology as almost one and the same thing, in the sense that the data just captures these psychological and behavioural biases. In other words, they’re just two sides of the same coin.”

One fund manager who has a particular interest in the whole science aspect and behavioural finance is Jupiter’s James Clunie.

The head of strategy, absolute return and fund manager of both of the group’s absolute return vehicles has for the past couple of years been the subject of a study by a professor from the University of Glasgow. It has involved his diarising various “passive” factors that may have some influence over the trades he is making. The thesis suggested that factors such as sleep, exercise, caffeine and alcohol intake, even the music one listens to on their way into work, could all play a role.

“Apparently sleep deprivation – anything less than six hours a day – can lead to poor decision-making and ultimately a bad trade. So I thought, as I’m so rarely sleep-deprived, why not give it a go? So every month I ship him data about me, then I send the orders we’ve placed and the prices we’ve got. And from that he makes his observations.”

Clunie says while gaining a better understanding of his distribution of which trades look good and which do not, if nothing else just doing the exercise has made him far more self-aware. “Are you getting enough sleep? Are you being fairly moderate in your diet? I’m sure you could take it to whole new layers and Fitbit yourself up or put yourself in a lab and take it further, but it is interesting. I find it makes you aware of any issues rather than seeing the evidence and choosing to ignore it.”

If one side of the propensity to ‘track’ behaviours and patterns around investment decisions lies in the backward-looking data and learning from it, somewhat ironically – given our industry’s constant reminders about past performance not being indicative of future performance – the other side lies in coaching.

Borrowing from the sporting world, businesses are now working in conjunction with leading athletic coaches. As well as identifying strengths and weaknesses, and building an appropriate coaching programme, other parallels exist.

Di Mascio says: “A lot of the work we do deals with the psychology of the investor. Ivan Lendl hasn’t suddenly made Andy Murray a better tennis player. We focus on things like motivation, coping strategies and resilience,
essentially. We have brought these things that are common in sports to our world, and we believe it’s making a phenomenal difference.”

While it may be straightforward to draw on psychological behaviours and understand inherent biases, we struggle to determine the precise fallout from our stress levels.

Janet Larsen, corporate psychologist and executive coach, points out how negative emotions can impinge on one’s cognitive capacity, so events such as an argument at work, an angry client, bad news flow about a company or something going on in the manager’s personal life will all cause emotional distress, impairing judgement, but importantly she notes  their effects will be residual.

She calls this a psychological “hangover” and points out it is therefore hard to overlay a timetable on to the after-effects: “It is important to be mindful of that and do everything one can to build emotional resilience, or grit.

“Coaching plays a role, both as a place to vent, work through, deal with some of the emotional fallout, but also to encourage behaviours, thought patterns and lifestyle that will underpin greater resilience,” she says.

While the value of monitoring past performances, observing tendencies and learning to improve certain behaviours is not in question – in the way a golfer might try to improve his swing, for instance – perhaps the parallels can only really be effective to a point.

As Professor Taffler points out: “When it comes to fund management, the opponent is not yourself or one person, or even a team of players. The opponent is the market, which is amorphous and uncertain.”

As humans, we are often guilty of over-engineering ourselves; focused so much on understanding, rationalising and justifying our behaviour in a bid to overcome our innate human traits, character flaws and striving for – ultimately, our version of perfection.

With the inundation and quality of data supply, plus the acceleration and reduced cost of technology presenting said data in a useable format, whether that information applies to stockmarkets, or indeed us, it is worthwhile.

While once the domain of the very richest of hedge funds and most innovative boutiques, as these approaches mature, costs come down and in turn access widens, the application of scientific disciplines to asset management should become the norm.

Clunie adds: “Active management is competitive and we should all try to be as good as we can be; there’s no finish line. You have to keep on learning all the way through.”