Author Topic: Notes in Nickel Steamroller Database vs Interest Radar vs LC's LoanStats.csv.  (Read 6389 times)

viking

  • Sr. Member
  • ****
  • Posts: 370
    • View Profile
    • Email
There appear to be some inconsistencies when comparing Notes in the Nickel Steamroller (NSR) database vs. Interest Radars (IR) database vs. Lendingclub's historical datafile (LoanStats.csv). For example, NSR finds a loan from 2008 that is not shown in IR's database and not readily found when filtering LoanStats.csv (e.g. using Excel).

 Try the following filter in NSR:

http://tinyurl.com/c4j4zqa

(it uses:
Credit Grade=G
Loan Length=36 months
Loan Purpose= Other
Inquiries=1
Total Credit Lines=57
Total Amount Funded=12,000
Year=2008)

Then compare this to filtering the data on IR:

Credit Grade=G
Loan Length=36 months
Loan Purpose= Other
Inquiries=1
Total Credit Lines=41-60 (includes 57)
Total Amount Funded=7-14K (includes 12K)
Year=NA (all)

Note that no Notes are found on Interest Radar.

Next if we filter the notes in LC's LoanStats.csv (e.g. use Excel and column filters):
There appears to be no loan with "Total Amount Funded"=12,000; only a loan with "Total Amount Funded"=35,000 (Loan ID 1208692)

The reason for this discrepancy is that the 12K loan is separated out from the rest of the loans in LoanStats.csv under the heading "Loans that do not meet the current credit policy". Nickel Steamroller includes these loans in the ROI calculation as described here: http://www.nickelsteamroller.com/blog/2011/10/current-policy-on-lending-club%E2%80%99s-data-export-%E2%80%93-what-is-its-true-meaning/. In contrast, it appears that they are not included in Interest Radar's database (unless I am missing something obvious)?

This should be considered when comparing ROIs calculated at Nickel Steamroller, Interest Radar and filtering the LoanStats.csv.
« Last Edit: November 25, 2012, 09:16:33 PM by viking »

NSR

  • Newbie
  • *
  • Posts: 35
    • View Profile
    • Nickel Steamroller
    • Email
Re: Incorrect Notes in Nickel Steamroller Database?
« Reply #1 on: November 25, 2012, 08:23:02 PM »
Pain staking hours have been spent on all software I make for public use on NSR.  I have spent this time to ensure NSR is reliable, accurate and robust.  I've leveraged years of software engineering experience, and personally invested hours upon hours testing and re-testing my products.  My database schemas have been designed from the ground up to prevent errors, duplicates and bad data from persisting. Additionally, I have data consistency checks that run 4 times a day. In fact, I even monitor for abnormal events in which for some reason a loan is no longer present in LC's export.

Not only this, but I include forums on every page of my site except for the main page to promote transparency. I aim to provide the best software on the web for P2P investors. I believe I am living up to this goal.

Please see loan ID: 267470

https://www.lendingclub.com/account/loanDetail.action?loan_id=267470

This meets of all the criteria of the loan you claim I have erroneously on my system. And indeed it is the loan NSR is correctly reporting.

Regards,

Michael

« Last Edit: November 25, 2012, 10:02:14 PM by NSR »

viking

  • Sr. Member
  • ****
  • Posts: 370
    • View Profile
    • Email
@Michael,
I am very sorry if I came across as making an accusation. This was definitely not my intent!
I love your site and think that it is the best free sites available for LC Return forecasting!

I was trying to understand why NSR showed a note that did not show up on Interest Radar and apparently not in LC's LoanStats.csv.
After your pointer to loan ID: 267470, I realized what was going on: You are including "Loans that do not meet the current credit policy" in the forecasting/ROI calculation.

I have rewritten my post above to clarify the reason for the discrepancies.

NSR

  • Newbie
  • *
  • Posts: 35
    • View Profile
    • Nickel Steamroller
    • Email
Viking

No problem. The original launch of NSR as you know did this include these loans, however it does now.

I appreciate you using my site, and I am very glad you find it useful. I'm always interested in working to make it better. If you have any recommendations or suggestions that would make it easier for you, never hesitate to make a feature request. You can do it right here in the forums if you like.

Regards,

Michael