Author Topic: Updates to your Lending Club data  (Read 13861 times)

jheizer

  • Sr. Member
  • ****
  • Posts: 472
    • View Profile
    • LC Tools
Re: Updates to your Lending Club data
« Reply #30 on: October 08, 2015, 12:42:54 AM »
It was in between drop times.  I happen to be on the web UI searching with and without exclude joint marked with no other filters just to see if any showed up and the count was toggling +/- 1.  I threw some debug code into my code but accidentally hit run after hitting a break point on it and lost the data as it was then gone.  I left it going the rest of the evening but never caught another one.  I don't save everything but this is yet another time I wish I stored all the loans always. 
Replacement to P2P Quant's Percentile Tool http://lc.geekminute.com

Fred93

  • Hero Member
  • *****
  • Posts: 2179
    • View Profile
Re: Updates to your Lending Club data
« Reply #31 on: October 08, 2015, 01:05:25 AM »
It was in between drop times.  I happen to be on the web UI ...

Grr.  We've never learned exactly why loans appear in between feeding times, but they occasionally do.  My sw runs periodically (every 2 hrs) between feeding times just to catch such stragglers, and also to handle the case where some money has popped into my account from payments, but of course that will only catch the ones that stick around awhile.

Quote
I threw some debug code into my code but accidentally hit run after hitting a break point on it and lost the data as it was then gone.  I left it going the rest of the evening but never caught another one.

I've been there.

jheizer

  • Sr. Member
  • ****
  • Posts: 472
    • View Profile
    • LC Tools
Re: Updates to your Lending Club data
« Reply #32 on: October 08, 2015, 01:11:37 AM »

Grr.  We've never learned exactly why loans appear in between feeding times, but they occasionally do.  My sw runs periodically (every 2 hrs) between feeding times just to catch such stragglers, and also to handle the case where some money has popped into my account from payments, but of course that will only catch the ones that stick around awhile.


I 200% agree. A few days ago I was changing some code and running it in production mode and it bought loans and I was completely shocked.  Previously I had decided that like you said, once the drop was done it was basically done.  I too now just do it every hour after that. Staying up to add in the code to record every loan now even if I never look at it again.  haha.
Replacement to P2P Quant's Percentile Tool http://lc.geekminute.com

Rob L

  • Hero Member
  • *****
  • Posts: 2082
    • View Profile
Re: Updates to your Lending Club data
« Reply #33 on: October 08, 2015, 09:34:12 AM »
If we have to keep guessing then the following from the Lend Academy Q&A with LC is helpful:

Q: Will the expected loss rate be lower given there are two parties on the loan?
A: Our experience would indicate that we will likely see a benefit in credit performance given there are two parties on the loan.
The ‘expected loss’ shown on the site will be based only on individual performance, until we see actual performance on joint applications at LC.

Another way of saying it is that the grade (interest rate) assigned to the loan will not change immediately.

I plan to ignore the new fields for now, the assumption being that a joint loan is never worse than an individual one.
If LC begins assigning lower interest rates to joint loans than would have been assigned an individual then my software will see them as less desirable. It may or may not buy them that being the case.

Fred93

  • Hero Member
  • *****
  • Posts: 2179
    • View Profile
Re: Updates to your Lending Club data
« Reply #34 on: October 08, 2015, 02:16:34 PM »
Here's one more tidbit from the lendacademy interview:
Quote
Q: What about the impact on your credit model? This seems like quite an adjustment for you and a good opportunity for investors.
A: The initial implementation of the joint credit model was built to be as similar as possible to our individual credit model, with the emphasis of the model on the primary applicant. You will see that the primary applicant meets all of today’s credit requirements (660 FICO, 3 years credit history, etc.) with the exception of DTI. People with high individual DTIs are in some cases given an increased loan amount or incrementally approved, where we find that the joint income is sufficient to cover the joint debts.

"emphasis... on the primary applicant"  Ok, but then note "with the exception of DTI".  Hmm.

We learn two things from this.  We see that they're thinking mostly about the "primary applicant". 

But they're not pure ... they are using the joint DTI.  (Without regard to the cosigner's FICO, etc, I wonder?)

Rob L

  • Hero Member
  • *****
  • Posts: 2082
    • View Profile
Re: Updates to your Lending Club data
« Reply #35 on: October 08, 2015, 04:13:11 PM »
But they're not pure ... they are using the joint DTI.  (Without regard to the cosigner's FICO, etc, I wonder?)

I wondered the same regarding co-applicant's FICO. LC could have thrown us a bone and included it as a new field too. At least we would have something to work with.
Clearly a 660 FICO applicant with a co-applicant that's 820 has gotta be very meaningful.
You might ask why not simply apply in the co-applicant's name in the first place, but not necessarily so. More like Mom or Dad "helping" in the background so to speak (or vice versa).
The 820 isn't likely to all allow his/her credit score to be trashed by late payments or a chargeoff.

PhilGD

  • Full Member
  • ***
  • Posts: 150
    • View Profile
    • Email
Re: Updates to your Lending Club data
« Reply #36 on: October 08, 2015, 04:16:37 PM »
If I were to guess, I'd say they won't show the co-applicant's score to us because maybe they're allowing co-applicants to have a credit score lower than 660, or even no credit history at all

Fred93

  • Hero Member
  • *****
  • Posts: 2179
    • View Profile
Re: Updates to your Lending Club data
« Reply #37 on: October 08, 2015, 05:14:15 PM »
I saw some joint loans appear at a regular feeding time...  So they are now in the regular flow.

10/08/15 14:00:03.2  LC 556 individual  3 joint  0 other

AnilG

  • Hero Member
  • *****
  • Posts: 1097
    • View Profile
    • PeerCube
Re: Updates to your Lending Club data
« Reply #38 on: October 08, 2015, 05:33:42 PM »
The two loans detected by PeerCube. The primary and joint annual income and DTI relationship is interesting. Is the DTI and Annual Income joint for the co-signer only or combined primary and co-signer together?

Code: [Select]
applicationType annual_inc annual_incJoint dti dtiJoint is_inc_v is_inc_vJoint
JOINT 25000.00 85000.00 37.11 11.13 NOT_VERIFIED NOT_VERIFIED
JOINT 48000.00 68000.00 26.08 20.42 NOT_VERIFIED NOT_VERIFIED

I saw some joint loans appear at a regular feeding time...  So they are now in the regular flow.

10/08/15 14:00:03.2  LC 556 individual  3 joint  0 other
---
Anil Gupta
PeerCube Thoughts blog https://www.peercube.com/blog
PeerCube https://www.peercube.com

Fred93

  • Hero Member
  • *****
  • Posts: 2179
    • View Profile
Re: Updates to your Lending Club data
« Reply #39 on: October 08, 2015, 05:45:45 PM »
The primary and joint annual income and DTI relationship is interesting. Is the DTI and Annual Income joint for the co-signer only or combined primary and co-signer together?

Great questions.  I've been presuming that "joint" means both taken together, but they don't actually anywhere explicitly say this.

Under my presumptions re what they mean, joint annual income should always be larger, but joint DTI could go in either direction, because I presume they consider the 2nd person's payments as well as his income. 

After I submitted my list of questions I realized that I was asking only about the meaning of the old fields in the new "JOINT" scenario, but actually there are things that are unclear about the new fields as well.  For example... The verification status of the primary borrower has three states: not verified, source verified, income verified.  Great.  Now I presume the cosigner also has three verification states.  That would mean that the joint verification should have nine states, right?  But they give us three states.  What have they done to reduce nine states to three?  If they took the minimum, then not verified may mean the primary guy is fully salary verified but the 2nd guy isn't.  On the other hand...


jheizer

  • Sr. Member
  • ****
  • Posts: 472
    • View Profile
    • LC Tools
Re: Updates to your Lending Club data
« Reply #40 on: October 08, 2015, 06:14:53 PM »
I only found 2 as well.  If you anyone wants to see the full data set.

{ "_id" : 62002798, "memberId" : 66195523, "term" : 36, "intRate" : "11.53", "expDefaultRate" : "4.65", "serviceFeeRate" : "0.85", "installment" : "395.89", "grade" : "B", "subGrade" : "B5", "empLength" : 72, "homeOwnership" : "MORTGAGE", "annualInc" : "48000.0", "isIncV" : "NOT_VERIFIED", "acceptD" : null, "expD" : "2015-10-22T13:53:36.000-07:00", "listD" : "2015-10-08T14:00:00.000-07:00", "creditPullD" : "2015-10-08T12:09:18.000-07:00", "reviewStatusD" : null, "reviewStatus" : "NOT_APPROVED", "desc" : null, "purpose" : "debt_consolidation", "addrZip" : "335xx", "addrState" : "FL", "investorCount" : 0, "ilsExpD" : "2015-10-08T14:00:00.000-07:00", "initialListStatus" : "F", "empTitle" : "Fire Protection Consultant", "accNowDelinq" : 0, "accOpenPast24Mths" : 4, "bcOpenToBuy" : 1191, "percentBcGt75" : "33.3", "bcUtil" : "84.7", "dti" : "26.08", "delinq2Yrs" : 0, "delinqAmnt" : "0.0", "earliestCrLine" : "2006-01-07T16:00:00.000-08:00", "ficoRangeLow" : 705, "ficoRangeHigh" : 709, "inqLast6Mths" : 0, "mthsSinceLastDelinq" : 0, "mthsSinceLastRecord" : 0, "mthsSinceRecentInq" : 0, "mthsSinceRecentRevolDelinq" : 0, "mthsSinceRecentBc" : 29, "mortAcc" : 1, "openAcc" : 19, "pubRec" : 0, "totalBalExMort" : 117522, "revolBal" : "6609.0", "revolUtil" : "24.4", "totalBcLimit" : 7800, "totalAcc" : 42, "totalIlHighCreditLimit" : 115794, "numRevAccts" : 13, "mthsSinceRecentBcDlq" : 0, "pubRecBankruptcies" : 0, "numAcctsEver120Ppd" : 0, "chargeoffWithin12Mths" : 0, "collections12MthsExMed" : 0, "taxLiens" : 0, "mthsSinceLastMajorDerog" : 0, "numSats" : 19, "numTlOpPast12m" : 1, "moSinRcntTl" : 11, "totHiCredLim" : 243046, "totCurBal" : 214899, "avgCurBal" : 11310, "numBcTl" : 6, "numActvBcTl" : 1, "numBcSats" : 3, "pctTlNvrDlq" : 100, "numTl90gDpd24m" : 0, "numTl30dpd" : 0, "numTl120dpd2m" : 0, "numIlTl" : 28, "moSinOldIlAcct" : 117, "numActvRevTl" : 1, "moSinOldRevTlOp" : 113, "moSinRcntRevTlOp" : 22, "totalRevHiLim" : 27100, "numRevTlBalGt0" : 1, "numOpRevTl" : 8, "totCollAmt" : 324, "fundedAmount" : "0.0", "loanAmount" : "12000.0", "applicationType" : "JOINT", "annualIncJoint" : "68000.0", "dtiJoint" : "20.42", "isIncVJoint" : "NOT_VERIFIED" }

{ "_id" : 62002793, "memberId" : 66195517, "term" : 36, "intRate" : "20.99", "expDefaultRate" : "11.8", "serviceFeeRate" : "1.11", "installment" : "753.4", "grade" : "E", "subGrade" : "E5", "empLength" : 60, "homeOwnership" : "OWN", "annualInc" : "25000.0", "isIncV" : "NOT_VERIFIED", "acceptD" : null, "expD" : "2015-10-22T12:14:59.000-07:00", "listD" : "2015-10-08T14:00:00.000-07:00", "creditPullD" : "2015-10-08T12:02:15.000-07:00", "reviewStatusD" : null, "reviewStatus" : "NOT_APPROVED", "desc" : null, "purpose" : "debt_consolidation", "addrZip" : "207xx", "addrState" : "MD", "investorCount" : 0, "ilsExpD" : "2015-10-08T14:00:00.000-07:00", "initialListStatus" : "F", "empTitle" : "cashier", "accNowDelinq" : 0, "accOpenPast24Mths" : 4, "bcOpenToBuy" : 8634, "percentBcGt75" : "60.0", "bcUtil" : "52.3", "dti" : "37.11", "delinq2Yrs" : 0, "delinqAmnt" : "0.0", "earliestCrLine" : "1996-09-07T17:00:00.000-07:00", "ficoRangeLow" : 690, "ficoRangeHigh" : 694, "inqLast6Mths" : 1, "mthsSinceLastDelinq" : 0, "mthsSinceLastRecord" : 0, "mthsSinceRecentInq" : 4, "mthsSinceRecentRevolDelinq" : 0, "mthsSinceRecentBc" : 9, "mortAcc" : 0, "openAcc" : 11, "pubRec" : 0, "totalBalExMort" : 39866, "revolBal" : "11542.0", "revolUtil" : "35.1", "totalBcLimit" : 18100, "totalAcc" : 18, "totalIlHighCreditLimit" : 28996, "numRevAccts" : 14, "mthsSinceRecentBcDlq" : 0, "pubRecBankruptcies" : 0, "numAcctsEver120Ppd" : 0, "chargeoffWithin12Mths" : 0, "collections12MthsExMed" : 0, "taxLiens" : 0, "mthsSinceLastMajorDerog" : 0, "numSats" : 11, "numTlOpPast12m" : 3, "moSinRcntTl" : 4, "totHiCredLim" : 61896, "totCurBal" : 39866, "avgCurBal" : 3624, "numBcTl" : 8, "numActvBcTl" : 4, "numBcSats" : 5, "pctTlNvrDlq" : 100, "numTl90gDpd24m" : 0, "numTl30dpd" : 0, "numTl120dpd2m" : 0, "numIlTl" : 4, "moSinOldIlAcct" : 85, "numActvRevTl" : 5, "moSinOldRevTlOp" : 229, "moSinRcntRevTlOp" : 9, "totalRevHiLim" : 32900, "numRevTlBalGt0" : 5, "numOpRevTl" : 10, "totCollAmt" : 0, "fundedAmount" : "0.0", "loanAmount" : "20000.0", "applicationType" : "JOINT", "annualIncJoint" : "85000.0", "dtiJoint" : "11.13", "isIncVJoint" : "NOT_VERIFIED" }




Edit: Oh I see they are both still available.  Will leave the details here anyway for future conversation.
« Last Edit: October 08, 2015, 06:24:51 PM by jheizer »
Replacement to P2P Quant's Percentile Tool http://lc.geekminute.com

jennrod12

  • Full Member
  • ***
  • Posts: 132
    • View Profile
    • Email
Re: Updates to your Lending Club data
« Reply #41 on: October 08, 2015, 07:02:23 PM »
Weird - if I search for the E5 loan on PeerCube, I can find it and view it in Lending Club.  But if I search for it on Lending Club with a filter that only narrows my selections to the E grade (no other filtering), it doesn't come up - any idea why?

Thanks,

Jenn

Prescott

  • Full Member
  • ***
  • Posts: 125
    • View Profile
Re: Updates to your Lending Club data
« Reply #42 on: October 08, 2015, 07:11:08 PM »
But they're not pure ... they are using the joint DTI.  (Without regard to the cosigner's FICO, etc, I wonder?)

Most likely they are doing "light" scoring on the co-applicant and using that light score to determine how much credit support the borrower gets. Ie if I qualify for $5000 solo, and my co-applicant is crap, maybe I only get $6000 offer, vs a really good co-applicant, maybe I get 10K offer

jennrod12

  • Full Member
  • ***
  • Posts: 132
    • View Profile
    • Email
Re: Updates to your Lending Club data
« Reply #43 on: October 08, 2015, 07:16:37 PM »
I filtered in LC on the B loan mentioned above and if I filtered on Dti of the individual (up to 30%) it found the loan.  But if I filtered on the Joint Dti (up to 25%), then it did not find the loan. 

My takeaway is that if you are using LC's basic UI to filter, you could miss out on a joint loan that might meet your desired criteria if the individual alone does not meet those criteria.

Jenn

AnilG

  • Hero Member
  • *****
  • Posts: 1097
    • View Profile
    • PeerCube
Re: Updates to your Lending Club data
« Reply #44 on: October 09, 2015, 01:50:04 PM »
Yep, Income verification field will be confusing if joint means both taken together. I noticed another JOINT loan today that seems to put real Lending Club incentive in perspective. The DTI of primary borrower is 59.19%. I guess JOINT application is to further expand the borrower base by loosening the credit criteria of the borrowers.



Great questions.  I've been presuming that "joint" means both taken together, but they don't actually anywhere explicitly say this.

Under my presumptions re what they mean, joint annual income should always be larger, but joint DTI could go in either direction, because I presume they consider the 2nd person's payments as well as his income. 

After I submitted my list of questions I realized that I was asking only about the meaning of the old fields in the new "JOINT" scenario, but actually there are things that are unclear about the new fields as well.  For example... The verification status of the primary borrower has three states: not verified, source verified, income verified.  Great.  Now I presume the cosigner also has three verification states.  That would mean that the joint verification should have nine states, right?  But they give us three states.  What have they done to reduce nine states to three?  If they took the minimum, then not verified may mean the primary guy is fully salary verified but the 2nd guy isn't.  On the other hand...
---
Anil Gupta
PeerCube Thoughts blog https://www.peercube.com/blog
PeerCube https://www.peercube.com