Do I have this right?

[Dennis]

I know this has been talked about on these boards at length, but do I have it right:

- notes at LC are issued en masse 4 times a day.
- the second (and I mean the absolute second) those notes appear on the platform, large institution/s with endless amounts of cash immediately suck most of them into their LC cache for review, temporarily locking them out to most small investors (the peer investors).
- then, after a relaxed review of those notes, the undesirable ones get tossed back onto the platform again, but often into the clutches of yet another institutional player - and the notes quickly disappear again.
- this process, where notes get pitched back and forth, gets repeated over and over again (in what seems like milliseconds), until they finally get fully funded.
- the small investor (the peer investor) must be quick and/or lucky to get one maybe two notes a session unless he/she is not picky on note quality.
- so the small investor (the true peer investor) who generally has limited amounts of cash, and maybe limited software capability, has very little or no chance of competing with all this.
- the small investor then is at the mercy of the big investor/s for notes, or for leftovers deemed too risky by the big guys to invest in.
- so the small investor (the true "peer" investor) starves while the big guys feast.
- P2P lending then is now mostly institutional lending, and we seem to have come full circle again in the finance community, where the big guys call the shots.  Nice concept, P2P - if it only worked.......

Now I know what I just wrote isn't always true or always happens, but I'm speaking in general terms where those things happen a lot or most of the time.

Did I get it right?

 

[brycemason]

As things stand today, this is more or less correct. I am less sure about a few points:

* I have no evidence that the large loan fractions which get tossed back are immediately carted by another large investor, but it could be possible that their software is scouring well past the load time.

* There is little evidence that the institutional investors are being intelligent in their selectivity. It would take an analysis comparing the fill time vs. an independent credit model score (such as Rev's IR04 or my PMax) to see if there is any correlation. Fill time is difficult to measure, although some on this board have made some attempts. If LC set their rates absolutely perfectly such that the expected returns across any loan grade were precisely equal, then the speed game would be pointless. You could invest in As and get the same return as investing in Fs. Unfortunately, I think there are legal / structural reasons why they can't use certain factors in their approval process and rate setting.

[standby]

I was going to say the exact same thing today - we have come full circle.  It can no longer be called peer-to-peer legitimately.  I think the big boys like to refer to it as micro lending.  I suppose that's because they were getting their foot in the door.  I've been here since just April and unfortunately, this looks like what it has become.  Why didn't they just go institutional in the first place I wonder?  Anyway, I don't know the exact details but it looks like you got the gist of it.

[mo]

- the second (and I mean the absolute second) those notes appear on the platform, large institution/s with endless amounts of cash immediately suck most of them into their LC cache for review, temporarily locking them out to most small investors (the peer investors).

It is actually more likely the technically inclined small investors who are doing this locking out.  Large institutions are probably using the official API which doesn't have the ability to lock a loan by adding it to your cart. 

[brycemason]

- the second (and I mean the absolute second) those notes appear on the platform, large institution/s with endless amounts of cash immediately suck most of them into their LC cache for review, temporarily locking them out to most small investors (the peer investors).

It is actually more likely the technically inclined small investors who are doing this locking out.  Large institutions are probably using the official API which doesn't have the ability to lock a loan by adding it to your cart.

This is unfortunately false, I believe. Day after day, I watch notes that I pick get funded. I am usually among the very first in the loan, and the "complete listing" page of the site says something like "$50 funded from 1 investor (me)." This page of the site only counts committed ORDERS. You can see what's been committed PLUS carted when you are viewing the loan in your shopping cart. So, when before I place my order, I'll see it "Closing," which means it's fully funded (through the carts). As I refresh the "complete listing" page, usually the very last investor to buy the fraction will make it jump thousands of dollars, or from 30% funded to 100% funded. Unless it's a very rich person, it's probably institutions using the carting-time-option.

[AnilG]

It is actually more likely the technically inclined small investors who are doing this locking out.  Large institutions are probably using the official API which doesn't have the ability to lock a loan by adding it to your cart.

Fully agree with this statement based on what I am seeing from loan updates on PeerCube. This can only be performed by people using browser or people who automated using browser/screen scraping. More and more it appears to be self-fulfilling prophecy. But in a way it is achieving what LC should be doing, spreading out availability of loans. Even 5 minutes after the scheduled release time, I am seeing enough new loans available.

8/29 8:35PM New loans added to DB = 11
8/29 6:20PM New loans added to DB = 13
8/29 6:05PM New loans added to DB = 58
8/29 3:05PM New loans added to DB = 18
8/29 2:35PM New loans added to DB = 6
8/29 2:20PM New loans added to DB = 17
8/29 2:05PM New loans added to DB = 78
8/29 1:05PM New loans added to DB = 1
8/29 12:05PM New loans added to DB = 1
8/29 10:21AM New loans added to DB = 9
8/29 10:06AM New loans added to DB = 62
8/29 07:51AM New loans added to DB = 1
8/29 06:35AM New loans added to DB = 12
8/29 06:20AM New loans added to DB = 21
8/29 06:05AM New loans added to DB = 61
8/29 05:35AM New loans added to DB = 3
8/29 02:05AM New loans added to DB = 8
8/29 01:50AM New loans added to DB = 1

[Dennis]

- the second (and I mean the absolute second) those notes appear on the platform, large institution/s with endless amounts of cash immediately suck most of them into their LC cache for review, temporarily locking them out to most small investors (the peer investors).

It is actually more likely the technically inclined small investors who are doing this locking out.  Large institutions are probably using the official API which doesn't have the ability to lock a loan by adding it to your cart.

One reason why I believe it's institutional investors locking these loans is that I've seen $35,000 notes disappear in that millisecond I talked about, but then several minutes later, the same notes reappear with only 20% or so funded.  It's almost unfathomable that enough small investors in tandem could lock out a note of that size so quickly, and then enough of them in tandem (again) reject it quickly from their cache where only 20% of it got funded - hope that makes sense.  I suppose it's possible, but seems unlikely.  It would seem more likely that if a note was rejected from multiple caches, where it eventually was only 20% funded, that it would first reappear on the platform at say, 98% or so, and then spiral steadily downward to that 20% as more and more small investors rejected that note. 

[Fred]

But in a way it is achieving what LC should be doing, spreading out availability of loans.

+1

I am glad to see this spreading out of loan availability.  Hopefully, this does not lead the bots to 'stay' busy 24-hr.

[brycemason]

Bots will work overtime if need be.

[Dennis]

Even 5 minutes after the scheduled release time, I am seeing enough new loans available.

I suppose it depends on how you define "available."  We could argue all day on these boards about what exact qualities make a perfect borrower as far as risk goes, but I don't think it would be hard to get a majority vote on the ones that truly suck, to put it mildly.  Those notes that make it to the 5 minute mark, they usually suck (sorry if that offends).  Not always, but most of the time.  I'm talking about the higher risk notes, D - G.  So to say they are available - technically, yes.  But to someone like me and I presume many others who care about note quality (again, I know we can argue all day about what that exactly is), those are not considered available notes.  No way in H. E. double toothpick I'd put my hard earned blood money into them.  I'm feeling a little raw tonight so sorry if the language is a little colorful.  And the thing is, some of those notes that you just know are going to default, or at least have about a 90% chance of doing so (based on very empirical data, empirical being defined in the philosophical sense - through experience), still get fully funded.  I some times wonder why investors want to lose their money like that, but I used to see it all the time as a stock trader.  There's ALWAYS someone willing to buy if the price is right (even if it's only in their mind).  Or put another way, there's always someone willing to invest if the conditions are right (even if it's only in their mind).       

[AnilG]

Bots will work overtime if need be.

Bots that filter and rank loans based on certain criteria will be defeated when instead of encountering 60 new loans it encounters 10 new loans every 5 minutes. 

It is not difficult to manage and control bots:

- Open up API for everyone like Prosper did.
- Throttle API bots
- Detect screen scraping and browser manipulation bots and time-lock the accounts being used by such bots

Being in IT infrastructure myself, I just wonder what kind of nightmare LC IT staff is going through because of a four-time-a-day loan release cycle. No amount of horsepower can manage the bot assault (if really happening) at release time. If it is really happening, it wouldn't be much different than being under DDoS attack and LC becoming susceptible to intrusion during release time.

[Fred]

Detect screen scraping and browser manipulation bots and time-lock the accounts being used by such bots

Anyone can put his browser behind a proxy server and watch the conversations between LC server and the browser.  With enough skills and diligence, anyone can design such program that LC server cannot tell the difference whether it is talking to a browser or to a bot.

[core]

Anyone can put his browser behind a proxy server and watch the conversations between LC server and the browser.  With enough skills and diligence, anyone can design such program that LC server cannot tell the difference whether it is talking to a browser or to a bot.

Agreed, there's absolutely no way to tell the difference.  If you could, there wouldn't be captchas and comment/forum spam all over the net. You don't need a man in the middle to do so, and with this stuff being HTTPS a proxy would be of limited use and/or more trouble than it's worth.  A freebie browser add-on acting as a crude sniffer is the only tool you'd need.

Any "solution" is going to have to involve the same rules for everybody, humans and bots alike.

[Fred]

If LC set their rates absolutely perfectly such that the expected returns across any loan grade were precisely equal, then the speed game would be pointless.

As an independent provider of P2P loan risk assessment, you certainly do not want this to happen, do you?  The pointlessness applies not only to speed, but also to advice. 

You could invest in As and get the same return as investing in Fs.

In its literal meaning, this statement would be counterintuitive.  Perhaps you mean that the returns of As and Fs loans are on the security market line (http://en.wikipedia.org/wiki/Security_market_line), rather than around it?

[Rob L]

Very interesting thread. I don't understand why a large investor would find it advantageous to lock up notes in a cart only to throw many back a short time later. Does this imply a human analyst in the middle? Why would their model / buy time be slower than Bryce's who said he often gets the first note?