I can’t seem to get
away from Lending Club. While working on the index measure in a previous post, I
had another idea. I think a couple of the test case ideas are pretty good. One
is the Zip/Grade case. This selects notes where past notes matching the zip code
area (3 digits) and the grade and sub-grade of the note have historically
returned more than 10% over the original purchase price. I found these were
hard to come by when searching for new notes, usually only about 2-3 showed up
each week.
I also feel the purchasing of notes on FOLIOfn where the
note is near the end and the borrower’s payment performance on that note is
pretty good is another way to select notes. It occurred to me one could
combine these two approaches and purchase notes on the secondary market with
the desirable Zip/Grades and a relatively short time to go might further refine
desirable notes.
I modified my program to look for this combination. I
learned a few things doing this. First, processing the exported available notes
file is a challenge due to its large size. But I could overcome this since I’ve
dealt with significantly large data files in the past. Second, the exported
file did not include the zip area. A tip
from user “AnilG” in the lendacadamy.com forum indicated the way to approach
this was to look at the past notes files from Lending Club. I then built a
cross-reference file of loan ID’s and the zip area. With this and my table of
returns by Zip/Grade I was able to select the desired notes.
This led to my 3rd discovery. Some notes are
listed more than once. I originally thought this was an error, but on further
review found that it wasn’t. While the data lines initially looked alike, the
asking price was different. Thus, different users are trying to sell the same
note, but are asking for different amounts. Makes you kind of wonder how multiple
investors decided to sell the same note near the same time. I’ll have to track
this. In any event, of the 350,000 available notes I only found 5 of interest
initially. So, I purchased theses and will put them in a new test case 8
portfolio. (I have subsequently added a couple more, but these are difficult to
find, so I’ll only be slowly adding these in the future.)
Here is the status of all the cases.
Description
|
Current
|
Other
|
Fully Paid
|
Charged Off
|
Index
|
Test 1 End in 2017
|
17%
|
0%
|
78%
|
4%
|
96.0
|
Test 2 End in 12 months
|
80%
|
3%
|
18%
|
0%
|
96.8
|
Test 3 LC Picks
|
97%
|
0%
|
3%
|
0%
|
99.6
|
Test 4 Zip Grade
|
90%
|
10%
|
0%
|
0%
|
100.0
|
Test 5 By Percent
|
90%
|
8%
|
3%
|
0%
|
99.7
|
Test 6 By Funded
|
90%
|
10%
|
0%
|
0%
|
100.0
|
Test 7 Random
|
93%
|
3%
|
5%
|
0%
|
98.6
|
Test 8 FOLIOfn Zip Grade
|
100%
|
0%
|
0%
|
0%
|
81.5
|
Again, it’s early so the Index isn’t overly significant yet
(particularly test case 8 because these notes are so new to me I haven’t
received any payments yet). I’m not sure when I’ll update my Lending Club stuff
again, but I’ll post on twitter when I do (@billlanke).
No comments:
Post a Comment