Mrs RT: Filtering the Currently Available Loans

Today's post is once again from Mrs. RT. As Mrs. RT became more interested in Lending Club loan selection process, she requested that I focus on automated loan filtering project for her to use instead of further analyzing historical loans.

Engaging in End-to-end Loan Selection Process

As Mr. RT is excited about gaining a fraction of interest every day through Lending Club, I’m becoming curious to know about the end-to-end loan selection process. Even though we select loans together at our Lending Club night, the prospective loans are pre-filtered by Mr. RT when I get involved in the final decision. It is like I’m taken to the restaurant Mr. RT has chosen and only activity I can do is selecting the food there. I want to see what kind of restaurants are out there and select what I really want to eat, too!

Exploring Lending Club Site

So, with Mr. RT’s account and password information, I explored the Lending Club to find loans for the first time, hoping I can find something that could lead more excitement to both of us. Once I logged in the site, first I found the double-digit net annualized return on the account summary page. Not bad considering the interest rate of 10 year treasury is less than 1.6% now and one year return of DJIA is less than 9%. I was skeptical about peer-to-peer lending when we started first, but I am convinced by the result. Keep up the good job, Mr. RT!

Next I started to browse notes. I found there were over 2,300 loans currently available for purchasing notes. I’m surely glad that Mr. RT pre-selects before Lending Club night. Otherwise, I’d be overwhelmed… But, wait, many filters are already available to select notes. I should be able to do it!

Trying Filters on LC Site

In one of my previous posts, I wrote a high interest rate loan would indicate borrower’s financial irresponsibility and I would avoid such loans as responsible lender. So, I selected interest rate A, B and C.

Now I think about double-digit return Mr. RT has achieved so far. It’s impossible to reach that rate if he selects A grade loans with interest rate of only 7.41%. From investment perspective, I excluded A grade borrowers. Now just over 1,300 notes have been selected.

Next let me exclude borrowers with public records and delinquencies last 2 years as they are considered to carry high risk of default. Even though I didn’t assume many borrowers would fall into these filters as they were categorized under higher level of credit score, to my surprise, more than 200 notes were excluded with these filters.Moreover, I totally want to remove borrowers with delinquencies. So, I chose 60 months or more since last delinquencies. There are still over 800 notes.

I want to make sure I’ll get my investment back! So, I selected borrower’s both debt-to-income ratio as low (max 20%) and revolving balance utilization as low (max 50%). Selected notes got down to almost 190.

Let me think… I’m sure Mr. RT has given valuable recommendations in his posts. Let’s see what he said.

He recommended 36 months over 60 months loans because of higher default rate of 60 months loans. So, I excluded 60 months term from the filter. Here’s another Mr. RT’s tip. He recommended avoid loans of extreme amounts. I set the max loan amount as $25,000 and still 130+ notes were selected.

Hmm, what other filters should I use…? Other filters do not seem reasonable to filter out loans. Employment length maybe, but the economy has not been good to people with even good credit score.

Discovering CSV Download

At that point I noticed a small icon at the bottom of the table, Download All. I clicked that and found all available notes, over 2.300, were imported as csv file. If I want to filter notes further, I have to filter the csv using the same criteria above again. It would have been better if only filtered notes through Browse Notes screen were downloaded at that point.

So, I started over my filtering using csv file. Most field names in the csv file are recognizable, but some are not. What are “percent_bc_gt_75” or “pub_rec_gt_100”?. It would be nice if Lending Club could provide the glossary of the description of each field for geeky lenders. Good thing about csv file is that now I have more flexibility to filter loans. I can filter out extreme low amount as well as per Mr. RT’s advice (less than $2,000), borrowers with any delinquencies in the past, etc.

As I’m interested to know about borrower’s financially responsible behavior, I filter out borrowers who opened more than five accounts in the past 24 months and had more than 2 mortgage accounts. At that point, I selected 70+ notes. There’s a field called “exp_default_rate”. Is it reliable data? I'm not sure, but I hate to lose money. So, I selected the lowest rate (2.3%).

Finally ...

Tah-dah! Finally 11 notes have been selected. It’s a good selection of notes I can recommend to Mr. RT and discuss. Mr. RT, let’s do Lending Club night!