Ahead of the AI & Big Data Hong Kong Leaders Summit 2019, we have an interview with David Ng. David joined CSOP Asset Management Limited as COO in 2017 is now responsible for designing and driving the infrastructure strategy of CSOP to support its business growth and development and is also currently leading the digital transformation initiative for CSOP.
What are your ideas for building data governance?
Data governance is a distinct concept of data management. It is a framework and strategy that could span a variety of processes, practices and disciplines, such as security, compliance, privacy laws, usability and integration. I am not a believer in taking a big bang approach or a one size fits all approach to building data governance. Organization should first examine the overall arching business strategy in determining the building blocks of its data governance. In conjunction with the business goals that need to be achieved, data governance requires determining what data can be used in what scenarios, which requires determining exactly what acceptable data is, how it is collected and used and the rules governing its collection and usage.
Data governance must also go beyond IT and include stakeholders from across an entire organization because the end-users of data is not IT but the entire business itself in its daily decision-making activities. Finally, data governance is a point in time concept, it must be an iterative process where we keep augmenting the framework to fit the business as we grow and as we consume a greater variety of data.
What are the main challenges for asset managers when enterprises become more ‘data-driven’?
These are some of the challenges I see.
As we become more data-driven, dealing with resistance to change and explaining to stakeholders that data-driven business processes can help in making better decisions and in some instances, outperforming human intuition and judgement. As we place reliance on data in our decision making, we also have to deal with the challenges of building a governance and control framework in ensuring the data we use are accurate and fit for purpose.
Portfolio managers and fund managers are increasingly using AI in their strategies nowadays. How effective is it without human oversight in asset management?
In its current state, I don’t think we are ready to have AI having full autonomy without human oversight. That would be a dangerous concept. There have been instances of blow-ups from algorithmic driven trading and investing. There needs to be a robust governance framework with human oversight on the design level for AI models, kill switches and thresholds must be set. There must be a regular review and stress testing of the model.
Do you believe that collecting much data is the cause of data swamp?
No, I don’t believe that’s the cause of a data swamp. The cause of data swamp is the lack of a data governance framework to manage the data. If there is a robust data governance framework in place, I don’t think there’s a limit to the data we can collect, subject to the confines of available technology to process that much data.
Big data means massive amounts of data, both unstructured and structured. Therefore, how do you turn your data swamp into the data lake?
At the design level, we cater to both unstructured and structured data. At the data layer, we utilize the traditional SQL-based databases such as MS SQL and NoSQL database technologies such as MongoDB. We utilize technologies like Hadoop to build data lake. At its heart, data lake is an architecture that consists of disparate components consisting of staging repositories where data are filtered and cleansed, search engines based on Big Data models and dictionaries, analytics modules and APIs for distribution of content via a multitude of UIs to various platforms.
David is a speaker at the upcoming AI & Big Data Leaders Summit Hong Kong 2019. To find out more about his presentation, please click here.