Big data makes its presence felt in small ways

London: Asking new employees to hand over urine samples for drug testing might seem distasteful to some, or alarming to others, but it has become relatively normal practice in corporate America.

This fact has not escaped the notice of one of the world’s biggest private equity companies, which has a majority stake in a large manufacturer of drug screening tests in the US. The private equity company found that analysing sales data from the drug-test manufacturer could help it build an early picture of whether US employment rates were going up or down: the more people being hired, the more drug screens sold.

This information could in turn be used to guide its investments in public equity markets.

This is just one example of how asset managers are beginning to realise the potential “big data” — analysing large amounts of information from previously inaccessible or overlooked sources — has for their investment processes, according to Craig Hapelt, a partner at Boston Consulting Group.

He says: “Everyone is analysing data, that is a given. But analysing it on a very wide scale and bringing in different pieces of data that have not been in place before, that is new.”

Although fund managers recognise that big data could provide valuable insights, many remain uncertain about how to apply vast data sets from sources as disparate as Twitter, the social networking service, Starbucks, the coffee chain, or Google Flu Trends, which analyses internet search terms to track global flu activity, to their investment processes.

Nick Thomas, a partner at Baillie Gifford, the Edinburgh-based fund house, says: “We are absolutely fascinated by the power of big data and how it makes Google a fantastic business, and we are some of the biggest holders of [companies] such as Amazon and Baidu, who are at the cutting edge of using data to build amazing internet businesses.

“But trying to fit some of those ideas into our investment process is difficult because we take a very long-term, slow approach to investing. We are thinking about it, but it is causing a lot of head scratching.”

Schroders’ head of investment, Peter Harrison, agrees asset managers need to get to grips with big data in order to remain competitive, but a spokesperson for the company said it has “nothing set in stone” on this front.

Harrison told FTfm in July: “Asset managers have got to get much smarter [and] work out how we can use the vast amount of data out there more effectively. Using data effectively will give you the vital, winning edge.”

Hapelt says he has worked with a large US pension fund and several sovereign wealth funds who are interested in how they can harness big data to improve their in-house investment processes. Many big investors have only started showing an interest in this area in the past 12 months.

The fact that fund groups have arguably been slow to realise the potential big data has to offer — both in terms of enhancing investment returns, and in understanding their own client base — has fuelled fears in the industry that internet companies with a better handle on data analytics, such as Facebook or Amazon, could easily enter asset management.

“I suspect that [Google’s artificial intelligence] people could clone an asset management stalwart before breakfast,” an FT reader recently wrote to the newspaper.

One branch of the asset management community ahead of the game is the macro hedge fund set that use a computer-driven investment process, according to Armando Gonzalez, chief executive of RavenPack, a data analytics company that works with more than half of the world’s biggest quantitative hedge funds.

Gonzalez says this section of the hedge fund industry has been purchasing his company’s analytics, which attempt to draw out market-moving company information from social media and news feeds, for several years, but that there has been a “significant” uptick in business from mainstream asset managers in the last two years.

The obvious drawback of using big data sourced from areas such as social media to influence investment decisions is that public sentiment cannot detect events such as insider trading, or substantial changes to monetary policy.

But it can raise red flags on a company when others are perhaps too optimistic, Gonzalez says. UK supermarket Tesco, for example, had a low score in RavenPack’s metrics (just 20 out of 100) over the summer. The score is based on factors such as shifts in analyst and social media sentiment towards a company.

This insight might have proved useful to investors such as Warren Buffett, who made the unusual move of stating he regretted his investment in Tesco after it emerged the retailer had overstated its forecast profits by 250 million pounds (Dh1.47 billion).

Deciding whether to change positions when big data flags up interesting patterns is the next challenge for fund groups, according to Michael Hintze, chief executive of CQS, the $14 billion multi-strategy asset management company.

He says: “One of the biggest challenges [in asset management] is how you filter and channel the ‘noise’ or ‘unstructured data’ and harness that. You also need the context and the ability to time the execution of the trade [based on that data] appropriately.”

Professor David Hand, chief scientific adviser at UK hedge fund group Winton, agrees: “There is more and more data captured every second. But data alone does not give you the answers. You have to ask the right questions [as] we are looking for very weak signals in a lot of noise. Only then can you attempt to predict how markets might behave.”

Winton’s research department has 140 people who specialise in data analysis and the company is “constantly hiring new people” in this area, according to Hand.

Making such a large commitment to assessing big data trends is a scary concept for more traditional asset managers, according to Hapelt.

“Some [fund groups] say [their ability to assess data] is such a mess, they can’t aspire to be any good. The hard part is [deciding whether to] invest in a whole load of stuff that might deliver in five years’ time, or might not get there. That is the fear.”

The consultant estimates that roughly 15 per cent of the asset management companies he works with have begun to organise their approach to big data to improve investment performance, and many are hiring specifically in this area.

To get on top of the opportunities big data can provide, Hapelt believes fund groups need to establish internally who is responsible for data management.

Appointing a “chief data officer” could be the first step.

— Financial Times