Print

[Fred93]

I published the updated chart in my blog tonite, but no reason for you to go there to see the chart, as its so easy to paste it here.

[patio11]

My quick eyeball suggests that things are getting better.  Perhaps not enough of an improvement to sustain Prosper's viability, but better.

Here's a quick scanning technique for you guys trying to make sense out of that.  Start with any line of interest at any point in time.  To compare that line to a loans of the same age, while stepping backwards in time, move one month back and look at the "previous" color, comparing its absolute reading to the line you are currently looking at.  So, for example, if I'm looking at the September 2007 line and want to see whether it is a quantum improvement over August, I'd look at September 2007 (currently at about 8% as of 2/01), shift back 1 month and scan for the peach line, which was at 10%.  Yep, things are looking better.

After you get a general feel for the pattern your eye needs to move in, you can do these sort of comparisons really fast.  Bottom line?  The changes wrought since last April are nothing short of amazing -- it is like watching two different products.  I'm thinking that is mostly a result of kicking out the HRs and rebalancing the risk pool as a result.  There is another discontinuity between August and September although the magnitude is much less -- what went down then, again?

I'd also be really, really interested in seeing a graph like this in just a single borrower pool of interest, like "clean Ds"

[Fred93]

The changes wrought since last April are nothing short of amazing -- it is like watching two different products.  I'm thinking that is mostly a result of kicking out the HRs and rebalancing the risk pool as a result.  There is another discontinuity between August and September although the magnitude is much less -- what went down then, again?

Most folks see two breaks.  First corresponds to when they raised the credit score threshold, chopping off the bottom of the HR ranks.  The second might correspond to the group changes or bidding guidance.

I'd also be really, really interested in seeing a graph like this in just a single borrower pool of interest, like "clean Ds"

Difficult.  When you break down the loans by origination month and also by some other criteria, you get not enough loans in each group, so you don't get smooth curves that you can make sense of.  There aren't very many "clean Ds". 

Somebody else publishes similar curves where he breaks 'em down by credit grade, but not by date of origination.  (His horizontal axis is just time from origination.)  This lets you see how the different grades have performed, but does not let you see whether newer vs older loans within those grades perform differently.  Sorry I don't remember who publishes those.  I don't even remember what he called them, so I can't search for 'em!  He used some word that means time passing, not "aging", not "seasoning".  Somebody will remember I'm sure.

[gelt4u]

Most folks see two breaks.  First corresponds to when they raised the credit score threshold, chopping off the bottom of the HR ranks.  The second might correspond to the group changes or bidding guidance.

I see the 2 breaks but I may see a 3rd.  It is way too early to tell, but 01/08 looks pretty.

[Urbi_et_Orbi]

It looks a bit better, but I think we should still let these loans season a bit more before we get all giddy.

[xraider]

Unless I'm missing something, the months starting 6/2/07, 7/2/07 and 12/2/07 don't appear to be notably better than what came before.

I'm with Urbi.... anything can happen.

Fred, thanks for posting your charts.

[Greebo]

My quick eyeball suggests that things are getting better.

Getting better would be indicated by the lines starting to go down.

[gelt4u]

My quick eyeball suggests that things are getting better.

Getting better would be indicated by the lines starting to go down.

Or by going up at a slower rate

[RateLadder_com]

I publish the "other" reports.  I call them vintage curves.  I will publish updated charts tomorrow.

[Mtnchick]

My quick eyeball suggests that things are getting better.

Getting better would be indicated by the lines starting to go down.

Or by going up at a slower rate

Losing money at a slower rate isn't something I'm after.

[gelt4u]

My quick eyeball suggests that things are getting better.

Getting better would be indicated by the lines starting to go down.

Or by going up at a slower rate

Losing money at a slower rate isn't something I'm after.

Oh come on now.  Stop being disingenuous.

To contemplate lending without considering losses is just silly.

So of course reduced losses would increase overall return.

geez

[Greebo]

My quick eyeball suggests that things are getting better.

Getting better would be indicated by the lines starting to go down.

Or by going up at a slower rate

Losing money at a slower rate isn't something I'm after.

Oh come on now.  Stop being disingenuous.

To contemplate lending without considering losses is just silly.

So of course reduced losses would increase overall return.

geez

Losses aren't reduced UNLESS loans start coming current.

Losses are simply increasing less quickly.

That's not improving.

[Cheerful_Lender]

Losses aren't reduced UNLESS loans start coming current.

Losses are simply increasing less quickly.

That's not improving.

We're discussing 2 different topics now...
1) Comparring one line to another (1 origination month vs another)
2) Comparring one dot to another on the same line (same origination month, but viewing it from different points in time)

I hope that clears up this debate.

[Fred93]

I publish the "other" reports.  I call them vintage curves.  I will publish updated charts tomorrow.

That's them!  I just couldn't think of the word "vintage".  Thought it was you.  Went to your blog and tried to search, but didn't know how to find them.

[traveler505]

Would it be possible to do a graph where the horizontal axis is calendar month of origin, the vertical axis is % delinquent, and the lines are two months after origination, three months after origination, etc.?  That would, I think, make it easier to visualize the trends in loan mortality as a function of evolving lender/Prosper behavior.