This section is only viewable on larger screen sizes. Please use a tablet or bigger. For best experience, please view on a desktop computer and use Google Chrome browser.

Can Artificial Intelligence be used to gain an investment edge in peer-to-peer lending?

Nominal Interest Rates by Grade Historical Returns by Grade The Gap Can We Do Better?
  • In this section we motivate our case study and approach.
  • Please click the links on the right to follow along.
  • Watch this space for analysis of the charts above.
  • To reset page, click on or near the page title.
  • P2P lending platforms bucket and grade borrower applications based on the borrowers' risk profile i.e. the predicted ability to repay the loan.
  • More than a 100 parameters are used to adjudge a borrower's application. These range from credit score (FICO) and debt-to-income ratio (DTI) to home ownership status, monthly income, and credit card utilization data.
  • A borrower's risk grade determines the interest rate on their loan.
  • We present IRRs for all loans made by a large P2P platform between 2007 and 2012.
  • Investors garnered annualized returns of 5% to 10%, generally with a higher return for higher risk.
  • Interestingly, returns for Grade F were a bit lower than Grade E, after adjusting for charge-offs, i.e. principal losses on defaulted loans.
  • While the gap between the APR on the loans and the IRR realized by investors increases with a worsening in grade, no grade is immune to the effects of charge-offs.
  • This chart serves as a baseline for our analysis, as we look to narrow this gap in returns, across grades.
  • Our goal is to utilize machine learning and AI to construct a superior portfolio that will outperform these baseline IRRs.
  • Here we show the Receiver Operating Characteristic (ROC) curve for various machine learning-based binary classifiers, ranging from a Deep Neural Network to a Gaussian Naive Bayes classifier.
  • The algos were trained and evaluated on a 1,000+ dimensional dataset. Among others, we use loan performance data with nearly 30 million rows.
  • Can we do better? The ROC curve points to a significant gain in information. The algorithms (with varying degrees of success) were able to tease out relationships and patterns in data that should help us create a more robust porfolio that outperforms the baseline.