Lot Fourteen collab creates new tool for banks to predict customer mortgage refinancing

Rick Rofe, founder and CEO of Neo Analytics and Aaron Lane, Principal Engineer at AIML

The financial services industry is among the most heavily regulated sectors in the global economy.

Strict regulations and compliance standards are mandated to ensure the safety of clients’ investments and compel institutions to manage their risk. These standards are heavily scrutinised by regulatory authorities and there are heavy penalties for those that fall short.

For banks, keeping up with compliance regulations and the continual changes that are made to them can place a huge drain on resources, particularly for smaller institutions. In fact, a recent Deloitte paper estimated that operating costs spent on compliance have increased by more than 60 percent for retail and corporate banks compared with pre-financial crisis spending levels.[1]

That’s where two Lot Fourteen businesses come in – Neo Analytics and the Australian Institute for Machine Learning (AIML).

Stone & Chalk Startup Hub resident Neo Analytics first started working with AIML as part of a State Government program that funded AIML’s engineering team to support SMEs looking to adopt machine learning technology to grow their business.

“It’s not uncommon for clients to complain that it can take them two weeks to prepare a monthly report the traditional way,” says Rick Rofe, founder and CEO of Neo Analytics.

“We saw the potential to create huge savings in time and resources if we were able to introduce a level of automation to the process of regulatory compliance and risk analysis.”

According to Aaron Lane, Principal Engineer at AIML, machine learning is about teaching computers to perform tasks by example instead of by rules.

“Much the way you would teach a person by showing them how to do a task and the results you would expect, we develop a scaffold for the machine to learn based on examples we provide” Aaron Lane, Principal Engineer at AIML.

“Machine learning is driven by data; you need it to be high quality and applicable to your problem. Institutions like financial services generate and manage large amounts of data that they are struggling to capitalise on. That’s where machine learning can make a huge difference.”

AIML engineers originally collaborated with Rick and his team to develop a machine learning model that uses a financial institution’s data to help predict when someone might be in danger of defaulting on a loan and therefore get in front of the potential problem. The solution, which took six months to develop, had to be thoroughly tested before the Australian Prudential Regulation Authority (APRA) sanctioned its use.

Even then, it had to be benchmarked against the traditional manual system. And the results were extraordinary, with the machine learning process turning a 3% “success” rate into a 60% positive result.

The two organisations then collaborated on a refinancing prediction model. While patterns are often not obvious when data is analysed manually, machine learning can help determine when and how often a customer downloads bank statements – which can indicate whether they are looking at mortgage relocation.

Neo Analytics’ Pliant platform has grown to include core banking mergers, compliance monitoring models, spend analytics, stress testing and provision monitoring to name a few. Their platform is now being used by institutions including Queensland Country Bank, Greater Bank and Beyond Bank. And the results continue to outperform the traditional methods by a factor of around 20 to one.

“The uses for machine learning in this sector are virtually limitless,” Rick said.

“The thing that we felt was really good is that we were taking the human out of the equation and the results we were getting were tailor made for machine learning applications. As soon as we started getting these results, our customers kept coming back to us for more. Consequently, we’re working with a number of organisations to move them from one system to another because of the machine learning capability we’ve built.

“The value we bring is that we integrate our products into the bank’s core banking processors – so once deployed, we’re not just solving a problem for one organisation but for multiple.”

For Neo Analytics, the Lot Fourteen ecosystem has delivered critical support as well as business opportunities. Being based at the vibrant and activated Startup Hub, managed by Stone & Chalk, has been a key driver in the company’s stronger focus on integrating machine learning into its products via its involvement with larger entities within the innovation district.

“The Stone & Chalk Investor Match and Mentoring Match programs have been really useful for us in terms of helping prepare our materials, fine-tuning our pitch, getting in front of investors and helping us learn about funding opportunities,” Rick said.

“There’s also another Lot Fourteen tenant that specialises in on-boarding new customers, and we’re now working with them to assist us.

“For a start-up like us, it’s brilliant to have such support and to be surrounded by a whole community of people who are innovating.”

Neo Analytics is growing at 60% year-on-year gross revenue and, just three years from launch, is now looking to South-East Asia and the USA for future growth. In fact, such is the potential that the company is starting to build its machine learning capability in-house.

For Aaron Lane and the 160+ members at AIML, that’s another win.

“While we do a lot of consulting to provide advice and direction, a big part of collaboration is about providing skills transfer and giving businesses the tools and knowledge to grow and bootstrap themselves in machine learning,” he said.

“When companies do that, it’s an indicator of success for us. As long as we’re helping South Australian businesses thrive and grow, then we’re doing our job properly.”