Peer to Peer Lending on Upstart

Our growth

Since launching our consumer loan product in May 2014, more than $6.9 billion in Upstart loans have been originated. In 2017, we experienced nearly 3X growth.

What’s driving this incredible growth? Our proprietary underwriting model goes beyond FICO scores -- it identifies high-quality borrowers based on signals of their potential, even if they have limited credit and/or employment experience. We call these consumers “future prime” borrowers.

The Upstart difference

We start with the same information other P2P lenders use (e.g., the applicant's FICO score, credit report, and current income). But we also consider educational and employment variables (such as degrees attained, school/university attended, area of study, occupation and employer) to develop a statistical model of the applicant's financial capacity and personal propensity to repay.

Our underwriting model leverages state-of-the-art machine learning techniques with thousands of data points to assign each borrower an APR based on their modeled likelihood of default. We verify each applicant’s identity, credit history, academic credentials, and employment using a combination of automatic and manual methods.

Repayments and Portfolio status

We model everything - repayments, delinquencies and prepayments of every loan, month by month. We continuously refine our model by comparing it to actual results. About 89% of Upstart loans are either current or paid in full.

A closer look at our borrowers

Weighted Average Fico © Score

Weighted Average Income

Refinancing Credit Cards*

* stated intent by borrowers, may not reflect actual use

These investors invest in people and earn solid returns

William Bunker

CEO, Clarity Health Services

“Investing on Upstart couldn’t be easier. I set up my account in minutes and let the platform do the rest.”

Jocelyn Ding

Vice President at Google

“Identifying quality borrowers with limited credit history is the holy grail of consumer finance. Leave it to a team from Google to find a data-driven solution to this age-old problem.”