Man vs. Machine

The growing role of machines in investment management

Read article and learn about:

  • The impact of robo-advisors on portfolio managers
  • The possible role of big data crunchers such as IBM’s Watson
  • Why machines will remain as decision-support for the foreseeable future
  • The future role of human fund managers

About the authors:

Clare Flynn Levy is Founder and CEO of Essentia Analytics, London, UK.Clare Flynn Levy is Founder and CEO of Essentia Analytics, London, UK.

 

Johan Bollen is Associate Professor at Indiana University and Founder of Guidewave ConsultingJohan Bollen is Associate Professor at Indiana University and Founder of Guidewave Consulting

 

Anders Quitzau is Innovation Executive and IBM Watson Advocate at IBMAnders Quitzau is Innovation Executive and IBM Watson Advocate at IBM

 


About the Panel

Clare Flynn Levy, founder and CEO of Essentia Analytics, has previously spent 10 years as a fund manager. Her fund management career included both active equity – running over US$1 billion of pension funds for Deutsche Asset Management – and hedge. Clare‘s vision for Essentia is based on years of being the client, followed by years of listening to the client.

Johan Bollen is Associate Professor at the Indiana University School of Informatics and Computing and Founder and CEO of Guidewave Consulting, a predictive analytics company. Johan Bollen has published more than 75 articles on computational social science, social media analytics, informetrics, and digital libraries.

Anders Quitzau a.o. a Masters in Finance and long practical experience from his jobs as management consultant. In his current job a Innovation Executive he is a Watson subject matter expert and enthusiastic advocate for the transformational new era of cognitive computing.


Joint interview with Johan Bollen, Clare Flynn Levy, and Anders Quitzau – three experts on new technologies that are affecting the investment management industry, including, but not limited to, social media analytics, behavioral bias, and big data. As they work to create new tools to support investment managers, they give their views on the growing role and power of machines, and explain what it may mean for the industry.

Journal:What’s the future of portfolio managers, taking into account the recent developments of robo-advisors?

JohanBollen: I’ve been stunned by the pace of change in this area over the past five years as a result of advances in artificial intelligence (AI) and machine learning. It’s incredible just how much decision-making has been (and is increasingly) outsourced to machines in areas that only a decade ago were deemed to be dominated by humans. I assume that this ‘encroachment’ from machines in supporting and taking over human decision-making will only increase in this area over the coming years. Robo-advisors in their current form are just the beginning.

Clare Flynn Levy: I can definitely also see where this trend is going, but think that in the near term there is a risk that we are getting slightly ahead of ourselves.

You can see it with the use of terminology such as ‘robo-advisors’ – as a term it is very misleading because there is no robot advising you. In most cases, a robo-advisor is simply asking clients a set of pre-determined questions to gauge their risk profile. Based on this, a human fund manager applies an appropriate investment methodology. So here it is important to note that humans are still very much operating ‘behind the scenes’ and are not being replaced by robots. So I sort of feel that the term ‘robo-advisor’ is more of a branding exercise, rather than an actual reflection of a robot actively managing a fund. We may get to a point where robots do control all investment decision-making, but we are definitely not there yet.

My company, Essentia Analytics, is currently working with a consortium of academic researchers on a project to build a system that can actually converse with a retail investor and understand their needs, and adapt to their requirements on an ongoing basis. Much like a human fund manager does today. This may be the future, but it is still a relatively long way off.

The future of machines in fund management is exciting and scary depending on who you are. The role of machines will no doubt increase, but today, the role of humans is still significant, and the current challenge is to find the right balance of how best humans can harness the available technology.

The role of machines will no doubt increase, but today, the role of humans is still significantClare Flynn Levy

Anders Quitzau: While I agree that humans will continue to play an important role, I do see a trend where robo-advisors are trying to ‘know’ clients better than they know themselves. In essence, going a step further than just gauging a risk profile. Having said that, it is yet to be seen whether robo-advisors can outperform human fund managers in generating returns, but my assumption would be that it’s just a matter of time.

...it is yet to be seen whether robo-advisors can outperform human fund managers in generating returns, but my assumption would be that it’s just a matter of time.Anders Quitzau

JohanBollen: I think Claire is right. Currently, machines can provide objective, quantifiable information that can support investment decision-making, but there is still a lot of room for humans. I think we’re both working on solutions to support human decision-making, rather than replace it altogether.

Journal: So we all agree that we can create decision support systems, but who will these systems support? The portfolio manager or the end users, i.e. private investors? Will it perhaps reduce the importance of the portfolio manager?

Clare Flynn Levy: An important question here is: Do peoplewantto make their own investment decisions? My sense is that people want to be in control to a certain extent, but ultimately, they don’t want to have to make those decisions on an ongoing basis? Research backs this up, suggesting that people fear regretting decisions they make and would often rather be able to blame a human (or a machine).

JohanBollen: We’re actually seeing similar issues with self-driving cars. The biggest challenge right now may not be the technology, but the ethical and liability issues. If something goes wrong, who makes the final decision and who’s held responsible for that decision? These same psychological and legal issues may also be troublesome in the investment management industry.

It’s also hard to predict how technology will factor into society. I’m noticing that the younger generation really appreciate good craftsmanship. If this translates to the investment management industry, it may actually mean that they will still want human involvement for the added value, rather than a soulless machine. That’s something that I guess only time will tell.

Clare Flynn Levy: That’s an interesting point. I think the first phase we will see, and are already seeing, is that technology will play the role of improving human decision-making processes. End investors have been demanding this for years. They want the industry to prove their added value.

JohanBollen:So at Guidewave Consulting, we are developing algorithms to analyze individuals, finding out their risk profile and tolerance. There are tremendous advances being made that allow machines to tap into a variety of signals, such as online behavior and text, and then draw conclusions from it.

Journal:We saw recently that IBM’s Watson just launched a headhunting service which analyzes your LinkedIn profile and can say which types of job you are best suited to. Is this just as possible with investors?

Anders Quitzau: Absolutely, and it will also be interesting to do that sort of profiling for thousands and thousands of people in order to help forecast market movements and the sentiment of the market.

JohanBollen: It will allow us to essentially tap into the psyche profile of lots of individuals and then see if there are correlations with market returns and fluctuations, like my research has showed with sentiment on Twitter.

We should remember that big data consists of lots of individually generated “small” data. When algorithms have access to longitudinal individual data, they can increase their accuracy by customizing and tuning their analytics to the individual instead of working with large-scale average patterns observed for large groups of people.

We’re going into an age where computers can learn a lot about us individually and translate this into more accurate individualized and collective analytics.

Journal:What are the key benefits of robo-advisors?

Clare Flynn Levy: I think their greatest advantage is that they start with a clean sheet, meaning that they have a lower cost-base and give much more attention to user experience. They’re creating a platform that is more adaptable to market demands. Robo-advisors have a lot of upside potential, but they cannot support any complex requirements, yet. Currently, they are only able to support the average person. A friend recently put it well, saying that robo-advisors have taken a lot ofmind share, but very littlemarket share. The reality is that people are lazy. And just like they don’t like to change banks, they don’t like to move their funds. This means a fairly substantial lag time for robo-advisors as they gradually build up their customer base.

A friend recently put it well, saying that robo-advisors have taken a lot of mind share, but very little market share.Clare Flynn Levy

JohanBollen: There’s a term in psychology: “cognitive overhead”. Even when the benefits of taking action outweigh the costs, if it requires people to consider a lot of new information and if it is mentally taxing, then people simply won’t bother.

Clare Flynn Levy: I can see it really taking off if people are ‘nudged’ by the government. Have it as an opt-out system, rather than an opt-in system. I feel like the adaption challenge isn’t technology; machine learning can advance in leaps and bounds, but if the process to move your existing funds to a robo-advisor isn’t super easy, then people just won’t do it.

Journal:Something we haven’t discussed yet is the potential role of robo-advisors supporting human fund managers.

Anders Quitzau: I understand that fund managers currently use 20-40% of their time on market research, 20-30% on meetings, and the rest on actually investing. There could be a huge opportunity for robo-advisors to help with the market research part, helping to find trends and make analysis and thus provide curated information for fund managers.

There could be a huge opportunity for robo-advisors to help with the market research part, helping to find trends and make analysis and thus provide curated information for fund managers.Anders Quitzau

JohanBollen: I couldn’t agree more, that is one of the strongest parts of the technology.

Clare Flynn Levy: I was on a recent roadshow with Kensho, an analytics firm with “the world’s first computational knowledge engine for the financial industry”. They are really disrupting Wall Street by pioneering real-time statistical computing systems and scalable analytics architectures. They are still mainly focused on selling to brokers at the moment, but it’s probably just a matter of time before they enter the buy-side market as well. Their software helps to understand market movements because it can find really interesting correlations, but at the end of the day, the person still needs to make the decisions.

JohanBollen: Machines are able to see through large amounts of information and find patterns and correlations as well as provide actionable intelligence. That’s something new. Machines can not only provide access to information, but also provide the intelligent analytics to help make sense of it.

Anders Quitzau: At IBM we have a project which is looking into whether machines can read a broad array of text and make sense of it. Could we for example, look at everything management of a company is saying in articles, blog posts, social media updates, etc., and find any interesting trends? That could be really useful information for investors.

JohanBollen: As I see it, the focus in the future will be on unstructured data. That’s where the value is. At the moment, the ability for algorithms to sieve through that data isn’t keeping pace with the amount of unstructured data being produced, but that is definitely where the future is.

As I see it, unstructured data is data that is waiting to be structured by a smart algorithm.

As I see it, unstructured data is data that is waiting to be structured by a smart algorithm.Johan Bollen

Clare Flynn Levy: The data we deal with at Essentia is not thatbig. We look at the data that is too big for humans to crunch, but not so big that it requires incredibly advanced technology. Even though all this advanced technology exists, and is making huge strides, the reality is that fund manager don’t even have the ability to take a moment to reflect on their investment decisions. So we believe that while there is a lot of great technology out there to support fund managers, the first step is to reflect on their own behavior, and then see which technologies could help them improve their performance.

Journal:So in summary, is it fair to conclude that you all see the future of machines in the investment management industry very much as a decision-support feature, rather than an all-out replacement of human fund managers?

Anders Quitzau:Yes, absolutely. There is still a need for human imagination, creativity and strategic insights to ‘outsmart’ the market, but no doubt their success will heavily depend on having access to the best robo-advisor in the market.

Clare Flynn Levy:Yes, in the near term the onus is on human fund managers to leverage machines to help them play to their strengths and avoid their weaknesses.

Johan Bollen:I agree, I believe machines can support not only the decision-making process itself, but will also generate the data and advanced analytics to allow humans to make better decisions.