lexo
Posts: 52 
Likes: 30
|
Post by lexo on Mar 12, 2019 16:31:38 GMT
On 1 March I pulled data from several platforms and run some analysis. I thought some people may find the comparison below interesting.
| ABL
| AC | FS | L | MT | Total portfolio
| £26,709,966
| £395,909,774
| £95,701,731
| £162,355,524
| £22,255,687
| Defaulted, % of total
| 11.6%
| 12.4%
| 59.3%
| 54.9%
| 29.6%
| Illiquid, % of total
| 45.7%
| 65.7%
| 23.8%
| 39.8%
| 47.6%
| Bad, % of total
| 57.3%
| 78.2%
| 83.1%
| 94.7%
| 77.2%
|
All numbers for FS include accumulated interests. Other platforms' numbers exclude interests. Bad=defulated+illiquid ABL: Defaulted: suspended/paused loans. Illiquid: loans that have not defaulted, but the best bid is 99.1% or lower (the minimum discount is 0.9% or higher). AC: Defaulted: suspended loans. Illiquid: loans with more than £20 000 units available. FS The standard term for each FS loan is 180 days. Once a loan more than 150 days old, the trading of this loan on the secondary market is automatically suspended. Defaulted: Defaulted loans + loans active more than 270(=180+90) days Illiquid: illiquid traded + illiquid non-traded where illiquid traded: loans on the secondary market with a discount of 0.9% or more; illiquid not traded: (loans active more than 150 days but fewer than 270 days)*(illiquid traded)/(active fewer than 151 days) where (illiquid traded)/(active fewer than 151 days) is a portion of all tradable loans that is illiquid. Loans active between150 and 270 days are not tradable, but they have not defaulted yet. The assumption here is that if the loans that are active between150 and 270 days were allowed to trade, the same percentage of them as of active loans would be illiquid. LY Defaulted: non-performing + claims underway. Illiquid: loans with more than £20 000 units available. MT Defaulted: defaulted loans. Illiquid: loans with more than £20 000 units available. Any feedback/suggestions/thoughts from fellow p2p investors is always welcome. Update: Two AC numbers are stricken-through because AC algorithm on the SM is the most different from other platforms. |
|
r00lish67
Member of DD Central
Posts: 2,692
Likes: 4,048
|
Post by r00lish67 on Mar 12, 2019 16:43:51 GMT
Any feedback/suggestions/thoughts from fellow p2p investors is always welcome. Interesting, thanks. First thing that leaps out is your definition of illiquid. One can still make a healthy profit trading out of loans at discounts of above 0.9%, and if they can sell at, say, 1.5% that's still liquid. So why term them illiquid for platforms with variable SM's? Also even if loans are totally illiquid for whatever reason, I don't see why they should be lumped into a category called 'bad'. For example, loans past <30 days on FS are illiquid because of the stipulations put upon them by the platform, but some would trade at par or even a premium if they were permitted to be. Or there might be a huge multi-tranche dev. loan on Assetz that is performing excellently, but has large availability due to its pure size - why should that suffer the label 'bad'?
|
|
benaj
Member of DD Central
N/A
Posts: 5,615
Likes: 1,740
|
Post by benaj on Mar 12, 2019 17:01:30 GMT
Hi lexo , that's interesting comparison. But I am confused about the definition of your representation. I've just logged in Lendy, The total loan amount (live+repaid+NPL+claims underway) issued exceeds 350+ Mil GBP |
|
lexo
Posts: 52
Likes: 30
|
Post by lexo on Mar 12, 2019 17:13:06 GMT
Iliqiud loans are simply the loans that most investors overall believe are riskier and, therefore, we can see either more units available or higher discounts. Those riskier loans are usually are harder to sell and hence the term "illiquid". I'm not saying that an investor should not invest in illiquid loans. Sure, you can make a profit there. Perhaps "less liquid" would be a better term.:-)
Similarly, "bad" loans are either loans that have already defaulted or the loans which the market believes will have a relatively high probability to default.
Those terms are used to have some comparison across different platforms to somehow compare the risk of loan portfolios of different platforms. Unfortunately, the definitions of illiquid and even defaulted loans are not consistent. Perhaps, someone can suggest a better and feasible way to compare risks on different platforms.
|
|
lexo
Posts: 52
Likes: 30
|
Post by lexo on Mar 12, 2019 17:15:24 GMT
loans past <30 days on FS are illiquid because of the stipulations put upon them by the platform, but some would trade at par or even a premium if they were permitted to be. FS is a bit tricky. I don't call all loans that between 150-270 days illiquid.
|
|
Greenwood2
Member of DD Central
Posts: 4,385
Likes: 2,784
|
Post by Greenwood2 on Mar 12, 2019 17:18:53 GMT
Looks terrible!
|
|
lexo
Posts: 52
Likes: 30
|
Post by lexo on Mar 12, 2019 17:19:30 GMT
there might be a huge multi-tranche dev. loan on Assetz that is performing excellently, but has large availability due to its pure size - why should that suffer the label 'bad'? Good point.
|
|
lexo
Posts: 52
Likes: 30
|
Post by lexo on Mar 12, 2019 17:21:40 GMT
Hi lexo , that's interesting comparison. But I am confused about the definition of your representation. I've just logged in Lendy, The total loan amount (live+repaid+NPL+claims underway) issued exceeds 350+ Mil GBP I don't take into account any repaid loans on any platforms. So for L it's live+NPL+claims underway
|
|
|
|
Post by mrclondon on Mar 12, 2019 17:23:16 GMT
lexo - the thing that immediately stands out is how you defined illiquid on AC (> £20k availability). The AC SM algorithm is not a queue in the conventional sense, but each purchaser gets a bit of each sale (with some non linear scaling in the background as far as I can tell). And most of the availability you see is DELIBERATELY engineered by AC via the QAA to be there to give people the availability. I more or less stopped lending on AC when there was minimal availability - if I can't get a 4 or 5 figure lump of a loan, I'm not going to spend the 4 to 6 hours doing the necessary due dilligence on the loan. I have restarted since they managed to get the availability up to make it worth my while.
I currently have parts for sale (at par) in a loan with £300k + availability, and they are selling down a few hundred a week. (I will be rebuying this loan in 3 weeks or so via next years ISA)
|
|
Balder
Member of DD Central
Posts: 646
Likes: 622
Member is Online
|
Post by Balder on Mar 12, 2019 17:27:25 GMT
on ABL paused isn't always a bad thing ie 100033 £650K is a 5 year loan not available for secondary trading.
|
|
lexo
Posts: 52
Likes: 30
|
Post by lexo on Mar 12, 2019 17:37:59 GMT
lexo - the thing that immediately stands out is how you defined illiquid on AC (> £20k availability). The AC SM algorithm is not a queue in the conventional sense, but each purchaser gets a bit of each sale (with some not linear scaling in the background as far as I can tell). And most of the availability you see is DELIBERATELY engineered by AC via the QAA to be there to give people the availability. I more or less stopped lending on AC when there was minimal availability - if I can't get a 4 or 5 figure lump of a loan, I'm not going to spend the 4 to 6 hours doing the necessary due dilligence on the loan. I have restarted since they managed to get the availability up to make it worth my while.
I currently have parts for sale (at par) in a loan with £300k + availability, and they are selling down a few hundred a week.
I understand that AC has a different algorithm. Still, some loans have 0 availability, while others have a large number of units available. The assumption here is that the former are more popular and the latter are less because overall they are deemed to be riskier. Would you agree? Surely, this may not work well perfectly for each loan.
|
|
lexo
Posts: 52
Likes: 30
|
Post by lexo on Mar 12, 2019 17:40:50 GMT
on ABL paused isn't always a bad thing ie 100033 £650K is a 5 year loan not available for secondary trading. There are always exceptions:-) Don't you think that if a loan is paused on AC it is usually a bad thing?
|
|
|
|
Post by mrclondon on Mar 12, 2019 17:54:26 GMT
lexo - the thing that immediately stands out is how you defined illiquid on AC (> £20k availability). The AC SM algorithm is not a queue in the conventional sense, but each purchaser gets a bit of each sale (with some not linear scaling in the background as far as I can tell). And most of the availability you see is DELIBERATELY engineered by AC via the QAA to be there to give people the availability. I more or less stopped lending on AC when there was minimal availability - if I can't get a 4 or 5 figure lump of a loan, I'm not going to spend the 4 to 6 hours doing the necessary due dilligence on the loan. I have restarted since they managed to get the availability up to make it worth my while.
I currently have parts for sale (at par) in a loan with £300k + availability, and they are selling down a few hundred a week.
I understand that AC has a different algorithm. Still, some loans have 0 availability, while others have a large number of units available. The assumption here is that the former are more popular and the latter are less because overall they are deemed to be riskier. Would you agree? Surely, this may not work well perfectly for each loan. No, AC availability has no (in most cases) relation to popularity or risk when viewed at the highest level.
Availability is primarily a function of the % of the loan that the QAA has retained since drawdown. The availability levels of loans don't change that much month on month because as units are bought by GBBA/MIA etc, further units are released to the market by the QAA.
The (new) loans that are currently offering discounts are those that have primarily been funded by underwriters not by QAA, these are the loans that will likely have low availability over the long term because the QAA will hold minimal amounts of them.
|
|
star dust
Member of DD Central
Posts: 2,998
Likes: 3,531
|
Post by star dust on Mar 12, 2019 18:03:20 GMT
lexo - the thing that immediately stands out is how you defined illiquid on AC (> £20k availability). The AC SM algorithm is not a queue in the conventional sense, but each purchaser gets a bit of each sale (with some not linear scaling in the background as far as I can tell). And most of the availability you see is DELIBERATELY engineered by AC via the QAA to be there to give people the availability. I more or less stopped lending on AC when there was minimal availability - if I can't get a 4 or 5 figure lump of a loan, I'm not going to spend the 4 to 6 hours doing the necessary due dilligence on the loan. I have restarted since they managed to get the availability up to make it worth my while.
I currently have parts for sale (at par) in a loan with £300k + availability, and they are selling down a few hundred a week.
I understand that AC has a different algorithm. Still, some loans have 0 availability, while others have a large number of units available. The assumption here is that the former are more popular and the latter are less because overall they are deemed to be riskier. Would you agree?Surely, this may not work well perfectly for each loan. I appreciate you're not asking me, but no I wouldn't - it can just be a sign of age and size of loan, the more recently released tend to have more parts available, in addition the lower the interest rate (therefore theoretically less risky) the less popular they tend to be, with the people who buy them directly on the MLIA. Don't forget a high proportion of most loans will be purchased by black box investors via algorithms. Underwriters can also have a big impact on loan availability on AC as indeed they are at present, the use of underwriters varies over time, and possibly says more about rate of growth of both the loan book and investors funds than it does about future or current liquidity or loan quality. AC is an odd beast that can be quiet difficult to categorise or predict in my view. on ABL paused isn't always a bad thing ie 100033 £650K is a 5 year loan not available for secondary trading. There are always exceptions:-) Don't you think that if a loan is paused on AC it is usually a bad thing? No - they are frequently paused for tranche draw-downs or capital repayments - you really need to look at the reason. It's possibly a bad omen for the majority - I've not done any analysis so really don't know - but certainly not all.
In Edit: X-posted with mrclondon - he obviously types faster than me 
|
|
|
|
Post by dan1 on Mar 12, 2019 19:22:22 GMT
... Any feedback/suggestions/thoughts from fellow p2p investors is always welcome. Hi lexo - thanks for sharing your stats, right up my street  It's probably not the same approach that I'd take if I had the chance but that's not to say there is no value in what you've posted, and the fact you state your assumptions helps us validate your figures (and my guess is that comments will come across as overly negative!). If it were me, I'd probably move the liquidity stuff into a separate analysis, being as liquidity in P2P is fickle at best and incredibly hard to assess. I'd probably stick to "performing" and "non-performing" loans rather than default and bad. You could break down the non-performing into formally defaulted, and/or 3/6/12 months overdue). (useful starter on NPLs: www.investopedia.com/terms/n/nonperformingloan.asp)I've looked at the FS figures to compare to my own analysis, some of which is posted here. Your portfolio figure of £95.7m doesn't look correct. My own figure is £84,966,124 on 1 Mar (prepared from all loans listed on the All past loans page plus a couple of others not listed), and that derived from the FS stats page is £85,191,134 (total lent - repaid - funds recovered - lost capital). FWIW I get 56.4% overdue by more than 90 days compared to your 59.3%, close enough!
|
|
It's probably not the same approach that I'd take if I had the chance but that's not to say there is no value in what you've posted, and the fact you state your assumptions helps us validate your figures (and my guess is that comments will come across as overly negative!).