Big data: what it means for investors

Analytical Surveyor has worked for over 40 years in the City and West End of London as a real estate investment manager, an equity analyst, a multi-manager, a fund manager and as a researcher and strategist. In his area of expertise, he is highly regarded and currently works as a consultant.

There is an enormous amount of academic research that provides insights into the functioning of investment markets. Rarely will this help with stock selection, but much of it will help in understanding how and why the market functions the way it does. The Analytical Surveyor trawls through the work that is available in the public domain and summarises the work into usable notes for investors. 

In this debut article for Master Investor, brought to you in partnership with Conkers3, the Analytical Surveyor explains how big data could provide a lifeline to fund managers desperately searching for alpha.

It’s a struggle being a professional equities fund manager. Most of the easy routes to market outperformance have been closed off many years ago: inside information, privileged access to broker analysts’ work, relatively cheap (compared to private investors) commissions, etc.

The ability to obtain market information early has been largely dissipated by the internet, making it available to the masses, hedge funds competing for the same opportunities, and high-speed traders acting much more quickly. In recent years, while the fund managers have costs to meet, including marketing, the academics are proving a case that average active managers behave averagely, but cost more than passive funds.

One potential saviour, of the big houses at least, is big data. This is defined as something like ‘extremely large data sets that may be analysed computationally to reveal patterns, trends, and associations, especially relating to human behaviour and interactions’.

(Note that this is not explicitly about generating the data, but about processing it. Although I do not profess to be an expert in the area of compliance, the distinction may be important for investors as, theoretically, the data is already available and, therefore, its mere processing should not imply ‘insider information’.)

Big data, big profits?

A recently published paper[1] by Zsolt Katona, Marcus O Painter, Panos N Patatoukas, and Jieyin Zeng seeks to demonstrate how big data can profitably be used in capital markets. High-resolution satellite imagery data of store-level parking lot traffic data is within the reach of sophisticated investors, with hedge funds being the typical clients of the providers.

The daily data feeds include point-in-time information about parking lot capacity – ie, the total number of available parking spaces – and utilisation at a specific time of the day. The raw data includes 4.8 million daily observations across 67,120 unique store locations for 44 major US retailers over the period from 2011 Q1 to 2017 Q4. The data covers 2,571 counties, representing over 98% of the US population.

This data was examined by the authors to determine if it provides timely insights for ‘nowcasting’ quarterly retailer performance. The evidence shows that there is a correlation between ‘parking lot fills’ and retailer performance but, more importantly, that the intelligence from the nowcasting data is incremental to existing intelligence and stock pricing indicators.


The authors find that a trading strategy that buys (short sells) retailers that experience an abnormal increase (decrease) in parking lot fill rates during the quarter generates abnormal returns over the three-day window centred on the earnings announcement date. More specifically, over the three-day earnings announcement window, the buy portfolio outperforms the market by 1.63% while the short-sell portfolio under-performs the market by -3.01%. The three-day spread between the buy and sell portfolios is 4.64%, which is statistically significant and economically important. This is even after accounting for the borrowing costs involved in short selling.

Short-selling is not generally available to private investors and, indeed, the authors also find evidence that while short sellers are targeting retailers with bad news for the quarter, individual investors are net buyers of such retailers. This raises the obvious question of unequal access to information and the ability to use it to the maximum extent.

There are, of course, margins of error in all of this, including that not all consumers arrive by car and there may be shared parking or undercover parking. It would, for instance, be difficult to extract the same intelligence from town centre traffic.

‘Unlimited’ potential

But the point is still good. The sheer computational power allied with large amounts of high-quality data can be demonstrated to provide the potential for out-performance. It is hardly surprising that it is hedge funds rather than institutional fund managers that are the main subscribers to this particular service, as they would have the ability to both place ‘big bets’ and to go short which, in itself, is not as risky as it once was in respect of the retail sector.

The same principles of satellite data acquisition could be applied to other outdoor activities including, for instance, house builders’ sites, estimating demand by the number of visits. Allying it with other data sources could also increase its potential.

Big data can be obtained about other activities, and, for instance, Google analyses the usage of it web search engine. Much of what it releases publicly is interesting but has little investment value – at least I assume that that is the case – but it, like the social media providers, is undoubtedly acquiring vast swathes of data about human activity and behaviour. Couple that to ‘artificial intelligence’ and there is unlimited potential to identify correlations, inflection pints, leading indicators, etc.

Virtually all of this data acquisition by players in the capital markets will be done privately, so not only will the results not be available in the marketplace but the mere knowledge of the usefulness of the data will be unknown. Unlike publicly-available data, the advantage of knowing this data could persist indefinitely. It is not, however, without cost, and that alone will limit some of its usefulness, but it will also limit its availability to those financial organisations that that have the resources to pay for it and to exploit it to its maximum. The smaller fund management houses may eventually see themselves squeezed between the private investors (with low costs and high flexibility) and the big players (with scale and resources). Unlike many of the so-called advances in technology, big data has a financial one and that means that it is here to stay and grow.

[1] On the Capital Market Consequences of Alternative Data: Evidence from Outer Space, 9th Miami Behavioral Finance Conference 2018 but posted on 15 March 2019, Zsolt Katona, Marcus O Painter, Panos N Patatoukas, and Jieyin Zeng