Print

[mothandrust]

I'd like to see these curves with the HR and NC data omitted. 

[PLP]

I'd like to see these curves with the HR and NC data omitted. 

+1   

[Fred93]

For all you guys who want to see the grades broken out, here is rateladder's latest "vintage curves" chart which he just posted on his blog.

[Mtnchick]

For all you guys who want to see the grades broken out, here is rateladder's latest "vintage curves" chart which he just posted on his blog.

You are so going to be in trouble for posting his copyrighted info. 

(thanks though, since I never read his blog )

[yankeefan]

For most credit levels, the size of the monthly "newly late" step is almost the same across the entire period- little advantage to having a "seasoned" portfolio-  12 month old HR's go late at the same rate as new HR's!

[RateLadder_com]

Please post a link... http://www.rateladder.com/2008/02/07/february-2008-vintage-curve-update/

[112233]

the image is a link to RL. it wasnt copied to here ..

[Tokyo Joe]

 

[HollowOak]

the image is a link to RL. it wasnt copied to here ..

I think a "link" means that one can click it and follow it to the source. Fred93 didn't post that type of link, he merely embedded the image in his post with image tags. That deprives Rateladder of a link that improves his search engine rankings.

I do think fair usage is that one should acknowledge the source of other's material with a link back to the original material in context.

[mothandrust]

My first question was why the the AA, A, and B lines drop off, but RateLadder explains that it's because there are less than 250 loans in the "bucket" after that point:

http://www.rateladder.com/2008/01/03/prosper-vintage-curve-update-112008/

I guess I look at Prosper as the same way I see Williams Sonoma or Pottery Barn--interesting products that I might buy if the price was reasonable, but at the price (rates) offered, I'll forever be a browser but not a buyer.

[Fred93]

My first question was why the the AA, A, and B lines drop off, but RateLadder explains that it's because there are less than 250 loans in the "bucket" after that point:

Yea.  Censoring based on size was a smart thing to do.  Notice how the curves are smooth toward the left of the chart, and get noisy toward the right?  That's because there are fewer and fewer loans on the right side.  For some of the curves there get to be so few that the thing just jerks around meaninglessly, so he censored those parts.

Let this be a lesson to those lenders who keep popping up with 20 or 30 loans who say they've got a great ROI.  The statistical noise is too high to allow measurement until they have over 100, and maybe even 200.

[epetroel]

Since there seems to be a lot of interest in this, I created something similar on ericscc.com (not live yet) but I wanted to give a preview here and get some feedback. 

I can do a line-type chart like Fred93's, but I thought another interesting aspect of this is to control for the loan age and look at how groups of loans performed when they were the same age.  An example:



This shows how loans from various months were performing after 6 months of aging.  The general trend appears to be downward - i.e. loans originated in Aug. of 07 were doing better after six months than loans originated in Dec. of 06 did after six months.  Whats more difficult to tell though is whether this is because lenders are picking better loans, or simply because the credit grade mix is different (e.g. more AAs and fewer HRs).  I suppose you could argue that these are the same thing, but I think it's more interesting to know if the average funded "C" or "D" loan today is better than the average funded "C" or "D" loan a year ago.  Unfortunately there just isn't enough data to draw much of a conclusion yet (I've filtered the chart by credit grade and the results are just all over the place). 

The actual URL for the image is:

http://www.ericscc.com/graphs/showLoanAgingGraph.php?age=6

If you change the "age=x" parameter you can view the chart for however many months of aging you wish.  Any feedback you guys can give me on what info I might add to that chart, filtering options, or any other data that I should show with the chart to help with analysis would be appreciated!

[arebelspy]

Here's 12 months: http://www.ericscc.com/graphs/showLoanAgingGraph.php?age=12

So from when Prosper opened and we first joined Feb 06 to the first PD in Feb 07.  Avg default looks to be about 20%.

Ouch.

[traveler505]

Here's 12 months: http://www.ericscc.com/graphs/showLoanAgingGraph.php?age=12

So from when Prosper opened and we first joined Feb 06 to the first PD in Feb 07.  Avg default looks to be about 20%.

Ouch.

There does seem to be a reduction in lates starting with loans funded in March 2007.  But the impact only seems to be a couple percentage points.  From other charts, I was postulating about a 5 percentage point drop, but I don't think we'll see it.

[ira01]

Here's 12 months: http://www.ericscc.com/graphs/showLoanAgingGraph.php?age=12

So from when Prosper opened and we first joined Feb 06 to the first PD in Feb 07.  Avg default looks to be about 20%.

Ouch.

There does seem to be a reduction in lates starting with loans funded in March 2007.  But the impact only seems to be a couple percentage points.  From other charts, I was postulating about a 5 percentage point drop, but I don't think we'll see it.

Isn't that when Prosper eliminated the NC's and bottom of the HR's?  If so, it is surely no surprise that lates were reduced since the borrowers most likely to be late were gone.