FMV of distressed notes

[william]

A bot is basically like a robot that does something on a computer. So for example if you have a web site that has ad you can download a bot that spend all day clicking on the ad (making you money). But this is how you use a bot for evil, in this case it would goes through loans and gathers info that can't be downloaded from LC like current employer.

I think I can find a couple of people to work on a bot like this, if nobody on here know/has the time to make one. But most likely they will want a "donation" or something.

The major problem would be getting the the urls of the loans
Example:
https://www.lendingclub.com/account/loanDetail.action?loan_id=1690172
https://www.lendingclub.com/account/loanPerf.action?loan_id=1690172&order_id=4222605&note_id=15084777

One way could be by using folio but that only has 5000-10,000 loans out of the 90,000+ that were issued.

Can anyone think of a solution to this?

[viking]

If anyone has collected information about *which* notes have gone into grace period, late 16-30, and late 30-120, I would be very pleased to do an analysis of them to find out any predictive factors that I can. It's too bad LC doesn't provide this variable in their extracts.

@brycemason,
William showed LC statistics which could be used to predict the probability that a note would be recovered.
What type of predictions do you think that you would be able to make if someone was able to get you the requested data?

[viking]

I base my prices off this chart and it has worked out pretty well for me.

So for example if a note is in grace period, I mark it down 16% not including folios fee and also not including accrued interest.
For everyday it's still on the market I mark it down 0.5% more. I try to get rid of a note when it's in grace period.
You also have to keep in mind the competition, so what I do is sort by loan id to find out what others are pricing the note at.

Would the same note be included multiple times in the statistics if:
case 1. It was late in several consecutive months?
case 2. It was late for 1-4 months, recovered and then was late again?

[william]

If anyone has collected information about *which* notes have gone into grace period, late 16-30, and late 30-120, I would be very pleased to do an analysis of them to find out any predictive factors that I can. It's too bad LC doesn't provide this variable in their extracts.

@brycemason,
William showed LC statistics which could be used to predict the probability that a note would be recovered.
What type of predictions do you think that you would be able to make if someone was able to get you the requested data?

The problem with what LC provides is this "How to read these graphs: of the loans in Grace Period in April 2011, 84% were partially or fully recovered by October 2011."

[viking]

If anyone has collected information about *which* notes have gone into grace period, late 16-30, and late 30-120, I would be very pleased to do an analysis of them to find out any predictive factors that I can. It's too bad LC doesn't provide this variable in their extracts.

@brycemason,
William showed LC statistics which could be used to predict the probability that a note would be recovered.
What type of predictions do you think that you would be able to make if someone was able to get you the requested data?

The problem with what LC provides is this "How to read these graphs: of the loans in Grace Period in April 2011, 84% were partially or fully recovered by October 2011."

So the data required would not only be if a note was ever late, but also how much was recovered and when. To complicate things, a note can be late multiple times, and that need also be considered. For example, a note that is late a 2nd time may be more likely (or not) to default than a note that was late the 1st time.....

[viking]

The major problem would be getting the the urls of the loans
Example:
https://www.lendingclub.com/account/loanDetail.action?loan_id=1690172
https://www.lendingclub.com/account/loanPerf.action?loan_id=1690172&order_id=4222605&note_id=15084777

The first URL, or the LoanDetails, is pretty easy to collect. Just download the history file from LC and loop through the Loan IDs.

However, the 2nd URL, LoanPerformance, require an order ID and Note ID because it shows the history of an individual investor's note. 
Simply using https://www.lendingclub.com/account/loanPerf.action?loan_id=1690172 doesn't work  .

[AmCap]

Viking - why would the second piece of information be necessary?  The payments on a loan are distributed to investors in pari passu right - so if you knew what the recovery was on the entire loan then you'd have all the information you'd' need.  Doesn't interest radar have that info for each note?

-

I'll have more on why I am interested in this information later - PM me if you are dying to know.

[yojoakak]

Viking - why would the second piece of information be necessary?  The payments on a loan are distributed to investors in pari passu right - so if you knew what the recovery was on the entire loan then you'd have all the information you'd' need.  Doesn't interest radar have that info for each note?

It's nice to see the Payment History, so you at least have some idea of what was going on with the loan.

For example, I've got "Late 31-120 days" loans that are making regular payments every month, but since they never quite catch up that information is completely lost at a high level. Here's one for example.

Plus, any Recoveries collected after a Charge Off only show up on the Loan Performance page. (Not 100% sure about this.)

[viking]

Viking - why would the second piece of information be necessary?  The payments on a loan are distributed to investors in pari passu right - so if you knew what the recovery was on the entire loan then you'd have all the information you'd' need.  Doesn't interest radar have that info for each note?

Because to calculate predictive factors, it may be important to understand what happened to the note after the initial late payment. For example, as suggested above, did it become late again?

[veggivet]

Calculating FMV rests on the assumption that you are dealing in an 'efficient market'. Does anybody on this board believe that the folio interface allows for this? All the calculations/bots you guys are discussing are, in my opinion, moot in this situation given the cumbersome construction of the interface on folio. 

[rawraw]

Calculating FMV rests on the assumption that you are dealing in an 'efficient market'. Does anybody on this board believe that the folio interface allows for this? All the calculations/bots you guys are discussing are, in my opinion, moot in this situation given the cumbersome construction of the interface on folio.

Which is a great opportunity for those who go through the effort.  For those who aren't familiar with the origins of option pricing and high frequency trading, you may find this podcast interesting: http://www.npr.org/blogs/money/2012/08/21/159373388/episode-396-a-father-of-high-speed-trading-thinks-we-should-slow-down  Although the FolioFN market isn't nearly as large as the options market, that guy became a billionaire off of figuring out how to price things before everyone else did.

But yes, I agree Folio has so many barriers I'm sure it isn't efficient.  And when I logged on last, it seems the YTM calculations are incorrect now.