Asset managers tap into big data

Asset managers tap into big data

  • Export:

From Google adWords to Amazon recommendations, big data is central to global e-commerce. Given these early successes, leaders in many other sectors are pouring money into developing data technology.

Big data, the collection and management of datasets too large and complex to analyse using traditional methods, was valued at $5.1bn in 2013 and is expected to grow to $53.4bn by 2017, according to PwC. Having recognised the potential of big data, asset managers are starting to develop ways to tap into its power. As a business built on information and analysis, asset management firms have always had to juggle large data sets.

However, over the past decade the data environment has changed substantially, with an explosion of both structured and, more often, unstructured data coming from both the financial industry and third-party sources. Demographic client details, news and commentary, blogs and social media posts, as well as data streamed from smartphones and other devices are all examples of data being created and stored at an extraordinary speed. Firms are now facing less a data pool than a data Niagara Falls.

The exponential global growth of big data, characterised by volume, velocity and variety on an unprecedented scale is fortunately matched by the development of high-performance computing technologies, from data processing to cloud storage. Simply put, enormous data sets can now be indexed and made searchable in a much shorter period of time, at a cost that makes utilising it viable. This has opened up bountiful opportunity for companies looking to gain information and insight.

But, as ever with technological developments, big data has created as many new challenges as it has opportunities. The main issue is one of implementation – how to turn big data into useful information and then real business advantage.

Chris Mills, UK director of consulting at Grant Thornton, says: “The holy grail is working out how to take the large data sets, which every asset manager has at their fingertips, and enhance them by layering on unstructured data such as social media.”

For asset managers trying to get ahead in an increasingly competitive market, information gleaned from big data represents a serious potential differentiator. Major firms, including Blackrock, JPMorgan and UBS, are reportedly making majorinvestments in big data technology.

Indeed, banks that have asset management and global custodian arms are in a prime position to leverage huge amounts of data gathered from third parties. State Street Global Exchange, a division that researches uses for its own data as well as developing data products for clients, was launched last year.

Securities lending data might be particularly promising to repackage as a stock-predicting indicator to sell to the wider market. Big data analysis techniques could also be used to identify the most difficult-tosource lending stocks in a more efficient way, which might bring more profit to the banks.

Searching for alpha

An unrelated State Street survey of more than 400 asset managers found that data and analytics was a strategic priority for 91% firms. Already, 86% had increased their investment in data and analytics over the past three years and two-thirds said that, in the future, data and analytics capabilities would be their most important competitive advantage.

Big data has already had some visible success in the development of indicators to advise stock picking. A number of social media indicators are now available through news vendors including Bloomberg and Thompson Reuters. These indicators gather social media posts relating to certain stocks or indices and use them to derive a sentiment indicator for each stock.

Chris Hammond director of the research signals division at Markit, says: “What particularly interests us is that it is an uncorrelated signal to the other factors in our library, which means it should be quite complementary to other factors and indicators that asset managers use.”

Released in April, Markit’s indicator uses raw data feeds purchased from Twitter via a company called Social Market Analytics. Only 10% of posts are kept, those which specifically reference the financial markets and are written by a user with a history of writing about stocks. From this selection the firm produces 22 indicators capturing sentiment levels, changes in sentiment and numbers of tweets.

Markit’s analysis found that once-a-day snapshots of the information can predict returns for up to five days. Stocks talked about positively showed cumulative returns of 76%, while stocks eliciting negative sentiment returned –14% over the two years to November 2013. Despite promising statistics, institutional investors are wary.
 
“It is still in the early stages of adoption. There is a lot of curiosity and everyone understands that there is some information in social media,” says Hammond. “What we have found is some of the most sophisticated hedge funds out there are looking at this sort of information. But, there is still a large portion of the investment community working out how sentiment data could be used for their investment process.”

A recent study by Warwick University found both Google and Wikipedia searches could predict falls in the stock market. The study found that there was a historical link between increases in searches about business or politics-related topics and stock market falls shortly afterwards. It back-tested trading strategies based on how often such terms were searched. These simple strategies produced statistically significant returns.

The mean return of strategies trading the S&P 500 using data on how frequently people Googled business-related terms between 2004 and 2012 was 37.4%. The mean return of the strategies using Google data on searches for politicsrelated terms for the same period was 56.4%.
 
Suzy Moat, assistant professor of behavioural science at Warwick Business School, says: “The logic was that Google has become a central source of information for many different people who frequently want to look for information to improve the quality of decisions they subsequently make.”

While not yet a useable indicator, the research has already gained the attention of the investment community. Tobias Preis, associate professor of behavioural science and finance at the university, says: “As you can imagine, this paper generated a lot of interest from institutional investors that want to develop these ideas further.”

Developing distribution

The other major avenue for the application of big data is in distribution. Interest has followed the great success of big data for distribution in the retail industry and new regulations encouraging asset managers to get closer to their clients. Big data gives sales teams better insight into their customers’ preferences, allowing them to decipher buying cycles and position products better.

Paul Mawson, investment management partner at KPMG, says: “The regulator in the UK at the moment is saying: ‘If you do not know your client, how can you be sure that you are not ultimately misselling?’ It probably is not the most comfortable place for the industry to be. Retail organisations that understand their customers particularly well can know who you are, understand your habits and then make relevant recommendations. That is an incredibly powerful service to be able to provide.”

Brian McCarthy, managing director of analytics strategy and innovation at Accenture, adds: “Assetmanagement firms can segment and analyse big data comprised of internal trade data, third-party data and social media data to create new investment products that would be of interest to customers amid changing demographics.”

Firms could also analyse their own social media data to gauge their advisers’ sentiment and apply that to business development. Finding new clients could be done through a similar approach by analysing a combination of a firms’ own social media data and external data to identify new prospects.

“Taking this a step further, a firm could combine its data and third-party credit bureau data with demographic and psychographic data to create a wealth index to help it better target new or existing clients,” says McCarthy.

More data, more problems

The potential of big data analytics touches almost every element of the asset management business, but seeking first-mover advantage presents significant challenges, not least the overwhelming issue of finding real value. Asset managers trying to create predictive functions will face large costs and their efforts will not necessarily produce results, especially those generating lasting benefits.

Paul Rowady, principal and director of data and analytics at Tabb Group, explains: “There is a low signal-to-noise ratio when it comes to finding short-term trading indicators through big data. It is questionable how powerful those are. You have to search through a ton of data to get a signal and then you do not have a history.

"It is still so new, you do not know how potent the indicator you think you have is. Those of us who watch closely realise how much data you need to go through and even then you run the risk of false positives. Long term though, there is a lot of value here.”

As with any advantage, big data insight may eventually be arbitraged away, sucking firms into a cycle of pouring increasing amounts of money into continuous development for little or no gain. Maintaining a competitive advantage will be an increasing problem, but there is hope says Jessica Donohue, senior managing director and head of research and advisory at State Street Global Exchange.

“The advantage gets reduced over time but how you deploy a new idea is a major factor in how productive it is,” she says. “Whenever a new data set gets revealed there is wide interest.

"Portfolio managers are generally sceptical that a new product is going to add incremental alpha, but more and more they are going to be pushed to look at new ideas and sources of data. Sentiment indicators, which are derived from data, were ignored until eight or 10 years ago, but today they are taken pretty seriously.”

One way that firms could deploy big data better is by combining external products with internal data, as its exclusive nature means any advantage it produces cannot be arbitraged away by competitors. It may be that external big data tools can get you to a point of opportunity, such as identifying a position on a stock, but translating that into action relies on the ability to pull data from across the business.

GoldenSource’s senior vice-president, product strategy, Stephen Engdahl says: “Selling or buying stocks is one way , but if you have a solid reference data foundation with good identifiers, you have the ability to look up all the different subsidiaries that those companies might have. You can identify potentially underpriced securities or optimally-priced securities to buy that might not just be the common stock of the parent.”

There are also some less technical but fundamental problems – a widespread shortage of staff trained to a high level in data analytics and a lack of well-defined regulation. Well-publicised cases have brought attention to privacy and ownership issues and, as regulation is in its infancy, the rules of the game could quickly change.

Donohue says: “At State Street we have a very respectful and conservative view of ownership of data. In any commercialised product we pursue consent . The question of data ownership is an important one that people are asking.”

New threats

While being quick off the mark contains risks, firms slow to improve data abilities might find themselves struggling to compete. This could be especially true if new competition emerges from non-traditional firms. Consultant Strategy& estimates that leading ?nancial ?rms risk losing 10% or more of their potential top-line revenue to non-?nancial competitors within the next few years if they do not move aggressively to transform and embrace innovation in data.

The success of Chinese technology firms, most notably Alibaba with its Yu’e Bao money market platform, entering asset management has accentuated the threat that new entrants with superior distribution can pose to traditional asset managers. One can only imagine the global distribution possibilities of Amazon or Google.

Tabb’s Rowady says data analytics capacity, particularly in distribution, is the most important advantage these new entrants would have, alongside user experience. “Firms that are born in the digital age are so much better at dealing with data analytics, finding patterns and delivering data. They are at the next level of pattern recognition in client behaviour. The incumbents are just not able to view the world like that.”

Beyond distribution, there is also the potential for large firms with a huge quantity of privately-owned data to deploy it in investment strategies to beat the market.

KPMG’s Mawson says: “Most of the new ways of finding alpha are arbitraging trends that other people have not seen. If you consider that information is everything in making these decisions, then there is a logical conclusion that says whoever has the most information and the ability to make these kinds of insights is sitting on a goldmine. Although the problem is the needle in the haystack scenario.”

Encouragingly for fund managers everywhere, Rowady believes asset managers have their own data-based advantage drawn from a long history of interacting with their clients, despite not managing this data very well.

As a result, he predicts increasing numbers of partnerships between technology firms and asset managers in the years to come: “Google, Facebook, Apple and other large technology firms have amazing amounts of valuable data but it is different from the type of data that the investment community have. They are capturing two different behaviours of the investing public, which is why ultimately the two groups could complement each other.

"You are going to see a lot of partnering going on because financial institutions realise they are not tech or data specialists.”

  • Export:

Related Articles