The Gateway to Algorithmic and Automated Trading

68% of financial professionals cite lack of agility in risk and financial model development as a fundamental business cost

First Published 14th January 2015

Collaboration between quants, developers, and business users is key to successful financial data analytics.

Steve Wilcockson, industry manager, MathWorks

Steve Wilcockson, industry manager, MathWorks

"Many see the benefits of implementing machine learning, a method often applied to big data, but the survey points to interest in more education, to understand the complexities of machine learning methods."

London - MathWorks, the developer of MATLAB and Simulink, has released the findings of a survey conducted amongst financial professionals from global organisations. The research highlights the shifting dynamics between data analytics and models. For example, it reveals that 68% of respondents identify "lack of agility to respond to market changes" as the biggest opportunity cost of slow model development. The full report is available to download from mathworks.co.uk/finance-modelling.

The survey also found that for 37%, risk management is the largest driver of model development and analytics within financial organisations, compared to just 15% who highlighted regulatory requirements.

Big data analytics is also driving change, but perhaps not as fast as anticipated with 46% of respondents working with gigabytes of data rather than terabytes or higher units. Adoption of machine learning methods, one approach to analysing big data, was limited to 12% of respondents, although 40% were keen to learn more about challenges, potential implications, and risks.

Overview of additional findings:

Model development

· 45% of buy-side respondents pointed to the potential for increased profit as a key driver; only 18% of those from the sell-side felt the same way.

· 37% of respondents identified risk management as the main driver behind model development within their organisations. Among the sell side, 54% identified risk as the primary driver.

· Just 15% identified response to regulatory requirements as the key driver for model development.

· Respondents report that a single software environment facilitates collaboration between model developers and business users (45%), reduces cost of model development (51%), and reduces risk (47%).

· 37% report that "creating effective models" is one of the biggest challenges associated with 'big data' in financial services.

Big Data and Machine Learning

· 46% of respondents are working with an average dataset of gigabytes, rather than terabytes or higher units. This compares with 49% in a similar survey conducted in 2012.

· 12% said their institution is already using machine learning.

· 45% are worried about issues such as over-fitting and a lack of transparency (black box approach).

· 40% felt they would like to better understand machine learning.

"Cultures and internal structures within financial services institutions are shifting. The increased focus on collaboration rather than the 'throw software over the wall' approach that has been commonplace for so many years, is consistent with the heightened awareness we are seeing regarding model quality," said Steve Wilcockson, financial services industry manager at MathWorks. "Agility, time to market, and quality are paramount - arguably the greatest weapons banks have in their bid to both hedge against risk and drive profit, saving them from the mistakes of the past. The findings suggest that despite the vast legislative change affecting the sector, regulation is not the sole or primary motivator for the industry's broader focus on risk management."

"There's a lot of enthusiasm for big data, but the survey highlights important themes where practitioners have concerns. Among the analytics challenges is appropriate model development and implementation alongside the data. Many see the benefits of implementing machine learning, a method often applied to big data, but the survey points to interest in more education, to understand the complexities of machine learning methods, so institutions can embrace opportunities without losing sight of the risks."