blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on May 31, 2014 15:53:23 GMT
There are a few changes on this page. We have lost the little note which told us how many of the live auctions were in our watchlist - an odd but long-standing feature. Recently this has enabled us to calculate the number of live 'whole loan' auctions, by subtracting the number of ordinary live loan requests. I think we were not intended to know that during trial and in future we are not to know what there is which we are unable to see. FC will keep their trial performance more private. Whole loans, by the way, now 24 in the loan book, just over £1M of exclusively A, B & C, and probably not much more to come.
The main change, however, and very welcome, is some graphs of actual % loss (bad debt) and expected loss for the various tranches of loans since the start, with months into the loan period as the x axis. This shows that the 2010 and 2011 loans diverged from expectation to produce twice the expected losses (6% v 3% after 30 months). But 2012 loans are currently conforming well to a raised expectation of loss. H1 2013 looks better than expectation so far, but it is early days and the C- loans do not yet figure in these results. The message seems to be that after a poor start, checking was improved and you might expect FC to keep improving both against expectations and absolutely. This may well be the case. Another message may be 'sell your old loans', which I do anyway. One thing that has not been corrected is that the average gross rate histogram still says it is for the last five loans when it is the last ten. To get 5 you have to first select a different number (try 10 - no change) and then select 5, to get the real histogram. Why has this not been fixed?
|
|
fasty
Member of DD Central
Posts: 1,038
Likes: 388
|
Post by fasty on May 31, 2014 17:42:48 GMT
The loss curves are very interesting. I was curious to see detailed wiggles and deviations in the expectation curves. (I would have imagined them to be smoother). Note the temporary levelling of loss rates after 14 months, 25 months and 27 months. What's this all about? And can we read into these curves that if a loan has survived just longer than a couple of years, then it's expected to be much more likely to survive the full course? Or indeed, if we sacrificed all our loans at the altar of the autobidder after 6 months, then we should have virtually no losses at all?
|
|
|
Post by elljay on Jun 1, 2014 7:49:27 GMT
There are a few changes on this page. Ah, that explains why my stats page broke...
|
|
ianb
Posts: 161
Likes: 54
|
Post by ianb on Jun 1, 2014 18:49:26 GMT
New stats page is an improvement but would be much better if by risk class so that we can validate the definitions. Ideal would be to dissect by different variables like region, industry type, term and so on. Likely Newcastle will win premier league first though.
|
|
pikestaff
Member of DD Central
Posts: 2,187
Likes: 1,546
|
Post by pikestaff on Jun 1, 2014 22:21:37 GMT
A more interesting bet would be whether Newcastle win Championship first
|
|
ianb
Posts: 161
Likes: 54
|
Post by ianb on Jun 2, 2014 5:46:33 GMT
100% more likely.
|
|
|
Post by GSV3MIaC on Jun 2, 2014 6:58:26 GMT
The loss curves are very interesting. I was curious to see detailed wiggles and deviations in the expectation curves. (I would have imagined them to be smoother). Note the temporary levelling of loss rates after 14 months, 25 months and 27 months. What's this all about? And can we read into these curves that if a loan has survived just longer than a couple of years, then it's expected to be much more likely to survive the full course? Or indeed, if we sacrificed all our loans at the altar of the autobidder after 6 months, then we should have virtually no losses at all?
You have to remember that the amount you have left to lose, 35 months into 36 month loan, is rather small, regardless of the chance it will go bust. The curve accounts for both those factors I imagine. What you need to know is % of your loaned capital that you lose each month (% of what is still loaned), which is a different curve.
|
|