Data Analytics as a Service in Asset Management: Moving from Assumption to Fact-based Modelling

The year is shaping up to be one where my dialog with leadership within large financial institutions shifts from “I want to do digital” to “Show me how digital helps me become more competitive.”

Digital has become table stakes and the asset management industry is now focused on how components within digital, such as data analytics, cloud or mobile, can create real solutions for customers and optimize the workplace.

Data analytics, for one, is at the cusp of moving beyond marketing and customer segmentation to the heart of what research analysts and portfolio managers do: move from assumption-based modelling to fact-based behavioural data capture. Ultimately, this transformation will give firms an edge on stock market performance predictions at a cost that is palatable and at a time when margin pressure is mounting.

Growth in social media, mobile, Internet of Things (IoT) and advances in technology to capture data (structured and unstructured) in the ‘micro-moments’ of customer interaction has created a vast trove of data often termed Big Data. The powerful combination of this data with the right algorithm correlation models, and team of experience industry professional can give asset management firms power to create fact based models. With platforms available today supported by teams of data scientists and industry experts, data analytics as a service is rapidly emerging in the asset management industry.

Moving to a Fact-based Prediction Model

Traditionally, firms have relied on smart people looking at company financials, market data and assumption models to forecast stock growth and other asset classes. These models depend on historical knowledge, financial engineering and public data to build predictions. The assumptions are then vetted, validated and fine-tuned by analysts tracking companies.

But what if there was a better way to leverage multiple streams of data, both structured and unstructured, in large data sets across social media alongside behavioural insights? What if there was a way to then build a correlation directly with stock performance combined with industry expertise, data analytics and behaviour signals from consumers and markets to create underlying assumptions for analyst models?

The option could potentially revolutionize research analyst models for buy-side firms and large financial institutions, with true predictive modelling based on fact signals instead of assumptions. By connecting real consumer behaviour with disparate internal and external data sources, you could create a method for making these signals both visible and actionable.

Asset managers are already using incredibly sophisticated analytics for clients, sales channels, advisors and to meet complex regulatory requirements. For example, JP Morgan recently won an award for its analytics platform, SPARTA, which includes real-time calculation of performance, contribution and attribution, in addition to on-the-fly grouping and advanced ex-post risk analytics.

With advances in AI and correlation models supported by data scientists, the platforms exist today in the market that provide tools set for next generation of investment management and research capabilities.

IDC, the global market research, analysis and advisory firm, predicts that the big data and business analytics market will grow from $130 billion to $203 billion by 2020. The banking industry is expected to be a big driver of this increase in spending.

Asset management companies, hedge funds and proprietary traders are investing to find innovative ways of using big data for investment research. The best of those will be closely guarded secrets but as an example of the lateral thinking being applied to data, Goldman Sachs is using satellite images of retail car parks as an indicator of retail sales both at an aggregate level and for individual companies.

That data forms part of the investment model and human judgment is still key to the investment decision. But by providing tools and clear signals to investment research teams so they can deliver correlations beyond financial data, firms can establish an edge over the competition.

Emergence of Data Analytics as a Service

With the current market dynamics, building a team of scarce talent, specifically highly skilled data scientists and deep industry experts who can filter signals from massive amount of data, is almost prohibitively expensive. So is the technology infrastructure to process and manage massive amounts of data across a wide variety of data collection points.

Even as the cost of data storage and process has shrunk considerably, each firm still requires a costly level of customization.  And if you go with free ware or open source information, the cost to build, maintain and really customize is still an expensive proposition.

Cloud computing has helped increase the speed and reduce the cost at which we can process structured and unstructured data. Multi-tenant platforms, which are customized to a firm’s specific needs for data collection and augmented by a team of expert data scientists and analytical people, can allow companies to cost-effectively leverage data analytics as a service.

A number of data analytics platforms which combine the expertise of data scientists, data collection and industry expertise while delivering a fact-based model of stock market and behavioural insights are emerging. These platforms offer a solution to asset management firms at a reasonable price point and limited capital investments to get started. Combining these platforms with in house expertise offers a viable Data Analytics as a Service Model to investment firms of all shapes and sizes.

Data used to be expensive, proprietary and the secret sauce for investment management firms. As information becomes more ubiquitous, data analytics and data science can give asset managers an edge by turning raw data, assumptions and customer behaviours into fact-based, actionable knowledge.

Data Analytics as a Service in Asset Management: Moving from Assumption to Fact-based Modelling

Leave a Reply

Your email address will not be published. Required fields are marked *