JSE slashes customer charges with new billing models, thanks to QlikView

In an era of relaxed exchange controls, the Johannesburg Stock Exchange (JSE) competes with many global household name exchanges, including the Nasdaq (US), Deutsche B"orse (Germany), NYSE Euronext, LSE (UK) and many more.

While it enjoys a good reputation as Africa's premier exchange, the exchange is under continuous pressure to modernise and compete in order to attract ever more trading activity from its members who act on behalf of investors.

“There is, for example, a generous R4 million overseas investment limit imposed on private individuals,” says Leanne Parsons, COO of the JSE and the head of equity investment. “Investors are free to pursue many investment options; we're just one of many.”

To achieve its objectives of growth and competitiveness, the exchange began investigating ways of charging less for trades in 2008. To prove the case for this switch in billing, it needed a tool to assess its impact on its business and customers.

“In the past we levied a flat rand-based charge per trade. We wanted to employ more sophisticated formulae that would offer greater choice and value,” Parsons explains. “To justify that, we needed a tool that could handle complex what-if analysis.”

What complicated the task of the tool was that the analysis contained many different variables on a per member basis - the size, the type and number of orders and trades, the share price and the discount threshold. “Analysis would be a multidimensional exercise,” Parsons says.

In addition, share information tends to change over time (due to, for example, mergers and acquisitions), and volumes are staggering. “We process hundreds of thousands of buy and sell orders every day, 251 days a year,” Parsons says. “The tool had to be able to analyse a billion rows of data, quickly and with ease of use, and it had to present the analysis flexibly.”

“Without a new billing model we couldn't grow the business in the way we wanted,” she adds.

“We needed a tool fast. Luckily, we were introduced to QlikView and liked what it offered: it seemed to do what we needed at a favourable cost, and so we went ahead with a proof of concept exercise. This was done on a representative set of data and confirmed that it was able to perform multidimensional analysis on massive data volumes, in memory.”

Parsons says the QlikView team was able to deliver on new requests within no more than two or three weeks at a time. Davide Hanan, QlikView SA MD, says this is due to the iterative development methodology employed by QlikView.

“Even if the user requirement isn't set in stone, this allows us to stage smaller development cycles and to build rather than re-write. It's a very agile way of arriving at the answer.”

In the end, the QlikView data model was delivered within a month - including testing - enabling the JSE to develop new billing structures.

“One of the biggest challenges in delivering a working QlikView data model was extracting the exchange's masses of data and understanding the way it is structured,” Hanan continues. Parsons says: “QlikView's prior experience with large, complex data sets in entities from all market segments proved to be invaluable in this instance.”

After doing an enormous amount of modelling throughout 2009, the JSE launched its new billing model to its customers in March 2010.

The new billing model, developed on the back of QlikView analytics, enabled the JSE to drop charges, Parsons says. “Instead of a flat rand rate per trade transaction, we were able to do more structured billing, including volume-based discounts, which has attracted more investment and is a powerful way for us to demonstrate our value to prospective customers.”

“In trades below R100 000, we were able to drop charges by close to 30%, while maintaining revenues and thus growing our business,” she adds.

QlikView also offers a quick and efficient visual analysis of JSE orders and trade data, as separate profiles. Running these huge queries on the JSE's production servers would negatively impact trading and downstream systems, creating a major risk. “The software's easy-to-understand graphical interface allows us to display and present close to real-time, saving additional time and resources,” Parsons adds.

“Having proved enormously successful in equity trading, QlikView has now stirred up a debate across the JSE about the enterprise-wide value of BI,” says Parsons.

For a start, the tool has entered the exchange's budgeting and forecasting division, and is being mooted for extension into settlement and clearance as well as other areas.

It is a powerful endorsement of the power of QlikView's powerful analytics to enable predictability across many scenarios for equity trading, and the tool is certain to unlock far more value across the organisation in time.