Grow your data science skills at posit::conf(2024)

August 12th-14th in Seattle

Funding Circle logo

Building a better financial world

Written by Marina Theodosiou, Risk Analytics Manager at Funding Circle

Funding Circle is the world’s leading marketplace exclusively focused on SME lending. More than £600m ($1bn) has been lent through its platform to more than 10,000 businesses in the UK and USA. SMEs can borrow directly from a diversified pool of investors, including more than 40,000 private individuals, the UK Government, local councils, universities and financial institutions.

Small and medium-sized enterprises (SMEs) account for more than 99% of companies operating in UK and US, providing two-thirds of all private sector jobs. Despite been significant contributors and drivers of job creation and economic growth, they have been traditionally underserved by financial institutions. A phenomenon that has exacerbated considerably during the recent economic downturn.

"We use Shiny to create an interactive environment between Underwriters and the Statistical Credit Risk Models. The establishment of interactive feedback loops through Shiny allows us to continually optimize our decision processes.”

Marina Theodosiou
Risk Analytics Manager at Funding Circle

The Challenge

Funding Circle was created to challenge the status quo and facilitate the financing of SMEs. Leveraging on financial technology it has built a track record in accurately pricing the risk of SME loans hosted on its platform, a challenge that traditional financial institutions have been reluctant to undertake. “At Funding Circle we saw an opportunity to improve the overall underwriting process for small business loans in a manner that’s transparent, with benefits to both lenders and borrowers.”

The Solution

There exist two main approaches to underwriting: automated versus manual, or statistical versus judgmental. Both these models of underwriting offer significant advantages.

Funding Circle leverages on both statistical and judgmental models and implements a hybrid semi-automated/semi-manual approach to underwriting.

Machine learning algorithms quantify the risk of default based on a number of factors, while manual underwriters review the model’s decision and complement the analysis carrying out an uncorrelated due-diligence on the business.

“We believe that the structure of our underwriting process and the way we optimally combine statistical and judgmental models have been the driving factors of our success.“

The Technology

To achieve an optimal combination of statistical and judgemental models to underwriting, Funding Circle introduced interactive analysis using the RStudio IDE for R and Shiny.

These tools have allowed Funding Circle to create an efficient communication channel between decision scientists/machine learning algorithms and underwriters and maintain collaborating feedback loops between the models.

“Through Shiny we are able to implement a virtuous cycle of analysis that is constantly improving our underwriting process, allowing us to serve small business effectively.”

With Shiny, underwriters can:

  • Update the inputs of the automated model based on the most up-to-date information they collect.
  • Use their experience to add insight into the model. Quantify and adjust the influence of contributing factors.

The Reward

RStudio and Shiny have allowed Funding Circle to achieve:

  • A “devless” environment where decision scientists have full control over the deployment of decision processes.
  • A testing environment where processes can be tested and monitored in real time.
  • Interactive feedback loops between underwriters and decision scientists.
  • Real time updating of decision process taking into account inputs from experience- based judgmental models

As a result, Funding Circle has built a track record of high quality underwriting of SME loans. It has gained the trust of thousands of retail and institutional investors who are willing to lend to SMEs through its platform, with immense benefits on employment and growth prospects in the economies in which it operates.


Subscribe to more inspiring open-source data science content.

We love to celebrate and help people do great science. By subscribing, you'll get alerted whenever we publish something new.