Author Topic: Can more filtering improve upon LC's expected returns in the long run?  (Read 14333 times)

Peter

  • Administrator
  • Hero Member
  • *****
  • Posts: 754
    • View Profile
    • Lend Academy
    • Email
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #15 on: April 24, 2013, 07:12:49 PM »
(Just to clarify, Peter: when you say "there would be no advantage in selecting loans - the returns would be uniform" I assume you mean selecting on criteria other than interest rates, since there is a legitimate reason why the returns for higher interest rate loans should be higher than for low interest loans, namely risk.)

Yes, I mean returns would be uniform within each loan grade if every note was priced perfectly.

And what zpbsfg says is true. Even though the vast majority of revenue is coming from originations I wouldn't be surprised if LC is actually making a loss or breaking even on that. Particularly today with them frantically trying to add new borrowers. So, I still maintain the main profit center going forward is going to be investor service fees.
Publisher of the Lend Academy blog

See my returns here: http://www.lendacademy.com/returns

Fred

  • Hero Member
  • *****
  • Posts: 1421
    • View Profile
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #16 on: April 24, 2013, 08:07:24 PM »
Particularly today with them frantically trying to add new borrowers.

Perhaps LC should change their current promotion strategy, from

Friends you invite receive $300 when they begin investing.

to

You receive $xxx when your friend's loan gets originated.

Zach

  • Administrator
  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Email
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #17 on: April 24, 2013, 09:24:23 PM »
Particularly today with them frantically trying to add new borrowers.

Perhaps LC should change their current promotion strategy, from

Friends you invite receive $300 when they begin investing.

to

You receive $xxx when your friend's loan gets originated.

That would be an excellent idea; considering their blog post about pausing active investor marketing, they should at a minimum temporarily discontinue that program.

Simon

  • Jr. Member
  • **
  • Posts: 51
    • View Profile
    • LendingMemo
    • Email
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #18 on: April 29, 2013, 03:24:24 PM »
As per the original question, I have been suspicious of this myself for some time.

It seems like those of us who statistically find weaknesses in the risk designation of loan grades are one step ahead of the platforms, and exploit this fine tuning to our own advantage and increased returns. IMHO, this will get harder and harder as the platforms assign their loan grades more and more accurately.

As previously stated, reduced risk in repeat Prosper borrowers is already becoming realized by the platform, as will the rest of our broad filtering categories. Income/EmploymentLength/etc.

However, as the bulky platforms take away the broad filters like income or repeat borrowing, the increased loan volume will mean we (as the lending community) can adjust our filters finer and finer. Increased loan history and available loans will mean we can have some really awesome very-fine-tuned filters in the coming years. So this should offset the platforms reworking their risk models.

For an analogy, think of big fish and little fish. Lending Club is like a fast great white shark that is eating up the best of the best (through loan-grade reevaluation and preference for institutional investors). However, us folks are like the barracuda! We will always just have to work harder, and the increased loan volume will make this harder work a fruitful endeavor.  8) Dexterity is the name of the game.
« Last Edit: April 29, 2013, 03:40:22 PM by Simon »
Writes at the peer to peer lending site LendingMemo.

MoneyTree

  • Newbie
  • *
  • Posts: 41
    • View Profile
    • Email
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #19 on: April 30, 2013, 03:55:04 AM »
I'm less optimistic about the long-term efficacy of filtering, even when it comes to small "fish", since unlike barracuda vs. white sharks we don't seem to have any intrinsic advantage/efficiency over LC/Prosper at detecting small opportunities. It seems to me that the only place we can hope for ongoing long-term advantage is through those parameters that LC and Prosper make public but do not themselves take advantage of due to legal restrictions. And even there, I can't help but wonder if it isn't just a matter of time until some borrower (or some attorney general?) successfully sues to prevent LC and Prosper from publishing those parameters, since doing so effectively aids and abets lending discrimination.

I really, really want to be wrong here, and I believe this is a very important issue. Somebody PLEASE provide me with a solid argument for believing that over time we will continue to be able to do a better job than LC and Prosper in filtering based on the same data they have access to.
« Last Edit: April 30, 2013, 04:41:20 AM by MoneyTree »

rawraw

  • Hero Member
  • *****
  • Posts: 2756
    • View Profile
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #20 on: April 30, 2013, 05:56:41 AM »
You worry too much.  Take a deep breath and you will know the definitive answer in time :)

Simon

  • Jr. Member
  • **
  • Posts: 51
    • View Profile
    • LendingMemo
    • Email
Re: Can more filtering improve upon LC's expected returns in the long run?
« Reply #21 on: April 30, 2013, 11:19:06 AM »
... we don't seem to have any intrinsic advantage/efficiency over LC/Prosper at detecting small opportunities.

I am looking at the richness of the historical data, and simply can't believe that platforms which issue $150 million in loans per month will ever find it a good use of their time to hyper-tune their loan grade assignments. Macro tune, yes. But micro tune? I don't think so. Concurrently, the richness of the historical data (and by this I mean both the number of columns and rows in Lending Club's CSV file) will also continue to grow, allowing us an increasingly growing ability to refine our targetting. Not only will additional borrowers be added per year, but the current platforms (as well as those yet to be launched) have an incentive to collect or verify more and more borrower attributes so as to better increase their measure of risk, and each new addition can rework our targetting.

The electronic medium of these borrower pools means consistent innovation for all involved. Peer to peer lending is dynamic by nature. (This frustrates us bloggers who find our online articles to already be out of date three months later, but se la vie). The beauty of consistent innovation is that it always benefits the small over the large.
Writes at the peer to peer lending site LendingMemo.

neals384

  • Full Member
  • ***
  • Posts: 107
    • View Profile
Well, I'm going to disagree with the notion that LC or Prosper should aim to create an efficient market: one where every listing is priced at the perfect balance between risk and return.

Actually, the platform goal should be to maximize loan funding.  The platform should try to price every listing at the right interest rate that entices enough lenders to fully fund it, while not pricing it so high that the borrower chooses not to proceed. 

To understand why the platform strategy is not the same as an efficient market strategy, consider the following example:  Suppose a large percentage of lenders refuse to lend to borrowers who are taking out Vacation loans (I know some of you have that filter).  The peer-to-peer platform then would, over time, discover that Vacation loans are being funded at lower levels, and their best strategy would be to adjust the underwriting so that Vacation loans, on average, have higher interest rates to attract more lenders.  Whether Vacation  loans, in fact, are riskier, is irrelevant to the platform.

Of course, if every lender is expert in evaluating the risk of each loan, then and only them does the platform's loan origination maximization strategy result in an efficient market.  Since that can never happen there will always be opportunities to beat the average with the right strategy.

MoneyTree

  • Newbie
  • *
  • Posts: 41
    • View Profile
    • Email
I am looking at the richness of the historical data, and simply can't believe that platforms which issue $150 million in loans per month will ever find it a good use of their time to hyper-tune their loan grade assignments.

Thanks for your thoughts, Simon. But if hyper-tuning is not efficient use of energy for the big guys, where the potential amount of money at stake is great, why would it be an efficient use of energy for the small guys, where far less is at stake? Unless the small guys view tuning as an enjoyable hobby, they too would be wasting their time. Actually, it seems to me likely that the big guys have more efficient means at their disposal than the small guys for fine tuning, thus giving them the competitive advantage there, too.

Simon

  • Jr. Member
  • **
  • Posts: 51
    • View Profile
    • LendingMemo
    • Email
Thanks for your comment. A couple of points:

1. Bryce Mason recently pointed something out: as long as their are interest buckets on Lending Club (B1, B2, etc), there will remain the ability to discover those on the safer "side" of each bucket class (in terms of lower risk). Until loan grades turn into a sliding scale, there will always be a place for filtering to discover the more rewarding edge of each bucket's gradient.

2. I'm not a statistician, but I think that whenever you have a large organization which is making generalizations about large bodies of data, even statistically accurate generalizations, there will remain a place for nimble analysis to discover vulnerability. Yes, Lending Club has better statistical tools than us hobbyists, but it has to direct those energies against a much larger situation; a hobbyist like myself simply has to find a few $50 notes per week. Furthermore, the math that we can use to analyze historical loan data is simply not that complicated; it is governed by the same basic laws as anything. The open nature of these platforms provides that benefit. In summary, I continue to believe that the electronic transparent and evolving nature of peer to peer lending will benefit retail lenders talented enough to do the work.

Would love for Bryce or Peter to wade in here.
« Last Edit: May 02, 2013, 01:03:27 AM by Simon »
Writes at the peer to peer lending site LendingMemo.

MoneyTree

  • Newbie
  • *
  • Posts: 41
    • View Profile
    • Email
Of course, if every lender is expert in evaluating the risk of each loan, then and only them does the platform's loan origination maximization strategy result in an efficient market.  Since that can never happen there will always be opportunities to beat the average with the right strategy.

Thanks for your thoughts Neals. As I understand it, your argument is that there will always be substantial opportunities to beat the market because the market will never be rational. I'm afraid I don't find that very compelling, particularly given the ever-increasing presence of institutional investors, who, presumably, are doing their best to be rational in their purchases. As more and more loan history is accumulated, the facts will slowly but surely reveal any such irrationalities to investors, and they will adjust.

Before I discovered LendStats I had many notions about what constituted a more or less risky note. It ends up that many of my strategies were actually counterproductive, and I have changed my strategies to become more rational. I have no idea what percent of LC notes are funded by individuals who are armed with nothing but intuition (and possibly-irrational notions) in making their investment decisions, but it seems likely that they are a decreasing percentage of total investors, thanks to sites like Lendacademy, lendstats, Interestradar, etc.

Yes, there will always be some irrational (uninformed) investors, but I have a hard time believing that their presence will result in an ongoing distortion in LCs underwriting practices that can be productively exploited.
« Last Edit: May 02, 2013, 01:06:07 AM by MoneyTree »

rawraw

  • Hero Member
  • *****
  • Posts: 2756
    • View Profile
These aren't equities.  It seems you are confusing the characteristics of an efficient equity market with loans (Fixed, contractual obligations) made to borrowers who live in a fluid environment.

MoneyTree

  • Newbie
  • *
  • Posts: 41
    • View Profile
    • Email
1. Bryce Mason recently pointed something out: as long as their are interest buckets on Lending Club (B1, B2, etc), there will remain the ability to discover those on the safer "side" of each bucket class (in terms of lower risk). Until loan grades turn into a sliding scale, there will always be a place for filtering to discover the more rewarding edge of each bucket's gradient.

Thanks again, Simon. Good point. I agree that as long as there are buckets, there will be an opportunity to exploit here.  Unfortunately, the opportunity is relatively meager; the buckets are all less than 1% away from each other, so, assuming all notes are properly rated, the maximum gain from such exploiting would be less than .5% at best.

(I wonder why the interest rate differences between the buckets are so irregular?)

MoneyTree

  • Newbie
  • *
  • Posts: 41
    • View Profile
    • Email
These aren't equities.  It seems you are confusing the characteristics of an efficient equity market with loans (Fixed, contractual obligations) made to borrowers who live in a fluid environment.

Thanks for your comment rawraw. I don't believe I am confusing the two, but perhaps I am. Would you care to explain why this should give us confidence that filtering will continue to yield significant gains over LC expected returns in the long run?

rawraw

  • Hero Member
  • *****
  • Posts: 2756
    • View Profile
For example, the impact of institution investors:

Quote
I'm afraid I don't find that very compelling, particularly given the ever-increasing presence of institutional investors, who, presumably, are doing their best to be rational in their purchases. As more and more loan history is accumulated, the facts will slowly but surely reveal any such irrationalities to investors, and they will adjust.

We aren't bidding on prices.  It doesn't matter if other people are irrational or rational.  Only impact it has is how long that note stays available for me to get in on it.   I just don't understand your concerns. Now Prosper 1.0 was a good example of how UNeffecient P2P was among retail investors lol