key capital logo

Happy Clients, Profitable Banks: The Possibilities in Machine Learning

It is always a pleasure to visit Denmark and with the battle on to become Nordic FinTech capital we were excited to

present at Copenhagen Fintech Innovation and Research (CFIR) recently on ‘Machine Learning and Financial Advice’

 

For those not familiar with the subject, ML is a form of artificial intelligence where the computer learns without being

explicitly programmed. It is extremely good for analysing huge volumes of data and variables, yielding insights that

humans simply cannot.

 

Moreover a computer is not constrained by the assumptions of statistics or human bias.:

 

 

Happy Clients, Profitable Banks

An explosion in the volume and complexity of data being produced, combined with the inability of old systems to process this, requires change. More importantly, it offers companies the possibility to gain competitive advantage through innovative new products, make personalised offers to clients, predict defaults and fraud, etc.

 

A recent study by Mckinsey on 12 European banks that replaced older statistical modelling approaches with ML techniques showed that, in some cases, they experienced a 10% increase in sales of new products, 20% saving in Capex, 20% increase in cash collections, and 20% decline in client churn.

 

A similar study by Cognizant on 537 corporations in N.America and Europe estimated that 45% of banks had seen 10% revenue growth from aligning analytics to front office activity. This was forecast to rise to 73% in the next 3-5yrs.

 

In a time when financial service providers face changes to their traditional business models these are compelling numbers.

 

 

Carbon v Silicon? It Depends!

In our presentation we looked at applications of ML in financial services, including Robo-advisory in treasury and wealth management, Telematics, AML and supply chain finance.

 

We saw that despite current benefits and future potential the speed of adoption has been slow compared to other industries. Why so?

 

Algo/managed trading has been commonplace for commoditised transactions in corporate treasury for years. Robo-advisors can automatically rebalance portfolios to keep them within the client’s pre-defined risk parameters.

 

Our client Union regularly automates processes like the entire loan life cycle. We know computers offer greater speed, consistency, auditability and cost efficiency but equally we understand the dilemma in delegating authority to a computer. It depends on the specific task and corporate culture.

 

 

Get Your IT Fit

Union's interest in ML is simple, its Digital Core and Credit Scoring solutions store the masses of data on which Big Data and Predictive Analytics can be based. At the moment they produce descriptive analytics about ‘what’ happened. Clients increasingly demand ‘why’ and ‘when will it happen again’. This is the key to unlocking client satisfaction and profitability yet misperceptions still persist:

 

1. Digital Transformation is a fashion. Nope, industries such as media and manufacturing demonstrate its power. Moreover regulation such as MiFID soon require active monitoring rather that static policies and historic data.

 

2. Digital Transformation can be achieved overnight. Nope, it as a journey with steps and milestones along the way.

 

3. Digital Transformation can be achieved on current infrastructure. Nope, you are going to need digitally enabled solutions – universal to eliminate silos, multi-channel, real-time, flexible, scalable.

 

 

Dont Wait

Navigating the transformation path when faced with a jigsaw of legacy systems, budget scarcity and high competition is not easy (or cheap!) but a sensible first step is process automation.

 

Short-term this allows data to be standardised, processes optimised and systems rationalised. Medium-term it creates the foundation on which to base analytics. Longer-term it supports the development of new offerings and capabilities.

 

Cognizant found that process automation was delivering significant savings to front, middle and back office operations, why not take the first step today!

 

 

p.s. TUSIND TAK to everyone at CFIR, it was a real pleasure and we hope to see you again soon.

 

machine learning in finance

11.08.2015