Free Updates

Exclusive content to your inbox for FREE!

The Super Simple High Return Strategy for Lending Club and Prosper

by Peter Renton on December 19, 2012

One of the goals of this blog is to teach investors how to get an above average return on their p2p lending investment. Lending Club claims their average investor return is in the 8-9% range and Prosper says theirs is in the 9-10% range.

In reality though, the average investor is probably earning a real return that is 1-2% less than those numbers. Now, 7% can seem like a great return today given the alternatives but I want you to do better than that. Much better.

So, I have been doing some research the last few days trying to come up with a very simple filtering strategy that has produced returns that are well above average over the last four years. I wanted to limit myself to just three criteria to keep it simple but at the same time I wanted there to be enough loans to choose from so investors can put their money to work quickly.

The Simple Lending Club Investment Strategy

Take a look at this graphic below from Lending Club’s site. This shows the average investor return, according to Lending Club, for each credit grade. The bottom line is that the easiest way to increase returns is to focus on the riskiest credit grades. This chart was taken from Lending Club’s site this week and is up to date as of December 7th. They have updated this chart regularly since it was introduced 18 months ago and it is always the lowest four credit grades that have produced the highest returns.

There is no guarantee that the future will produce similar results and I should point out that it is more risky focusing on the lower grade loans but it has historically been the easiest way to produce above average returns.

Let’s cut to the chase. Here are the three criteria I have chosen for a simple investment strategy that has produced above average returns in the past.

  1. Credit grades D, E, F and G
  2. Inquiries in the last 6 months = 0
  3. Loan Purpose: credit cards, debt consolidation

The question you probably want to know is how have these loans performed historically. The table below provides the details. I broke up the returns into two segments: 2009-2010 and 2011-2012. I did this because I wanted to see how this strategy has changed over time. The key number is in the last column – that is the percentage above the average return. If you had employed this strategy over the last four years you would be 3.5% – 4% above the average investor.

YearsAll loansAll loans ROI3 Filters Loans3 Filters ROI% Above Average
2009-1017,8155.64%1,0699.22%3.58%
2011-1272,5868.42%3,30512.33%3.91%

You can see the most up to date returns for these criteria on Nickel Steamroller (where I did my research) – click here for the 2009-10 returns and here for the 2011-12 returns.

The Simple Prosper.com Investment Strategy

At Prosper there is a similar trend for credit grades. The bottom four credit grades have historically provided the best performance, at least for Prosper 2.0 loans. The chart below shows seasoned returns, meaning it only takes into account those loans that are at least ten months old. It is up to date as of September 30, 2012.

Long time readers may be surprised by the three criteria I have chosen for Prosper – it doesn’t include one of my historically favorite groups: repeat borrowers. While repeat borrowers have provided excellent returns in the past I have noticed a recent downward trend which is why they didn’t make the cut for my three criteria. Here are my three criteria that have consistently performed well.

  1. Credit grades C, D, E and HR
  2. Inquiries in the last 6 months = 0
  3. Open credit lines >= 10

So, basically we have very similar criteria to Lending Club, just swapping out loan purpose for number of open credit lines. But the difference in returns at Prosper is even more impressive than at Lending Club. We are talking returns of 5.5% – 7% above average and with plenty of loans to choose from. Here is the table showing the breakdown.

YearsAll loansAll loans ROI3 Filters Loans3 Filters ROI% Above Average
2009-107,6899.01%60016.06%7.05%
2011-1230,3188.76%2,34314.39%5.63%

I produced these numbers from Prosper Stats – you can also follow these links and see how they perform over time. Click here for the 2009-10 returns and here for the 2011-12 returns.

Words of Warning

First, I should point out again that investing in the lower credit grades is inherently risky and not for everyone. You will receive many defaults even with the suggested filtering. And if we do go over the fiscal cliff and fall into a deep recession again then these lower grade loans will likely be hit the hardest. So, there is no guarantee that they will produce above average returns going forward.

Second, as I say in my disclaimer the information in this post should not be construed as investment advice – I have no formal qualifications so I am not able to provide such advice.

So now I turn it over to you. What do you think? Are their better ways to generate above average returns using just three filters? As always I like to hear what you think.

{ 52 comments… read them below or add one }

Michael December 19, 2012 at 12:01 pm

Peter, I’ve been using a identical strategy for over a year now with 12+% and 16+% returns for Lending Club and Prosper respectively. I do find the number of loans to be limited at times so I have also included auto & home loans too when I find new money not funding at the pace I would like. After I have money invested and I want to go into a mode where I am just minimizing cash drag I will removed auto and home loans from the filters. (FYI All of my portfolio fall into the 2011-2012 range for age)

Debt loans are still the best bet for high returns.

I’ve seen really complex filters that produce returns over 16% on LC but have less than 100 loans for the entire platform. I think this is fine for people investing small amount of money but if you plan to put in 10k+ this post is an outstanding starting point and should be a staple of any portfolio. Additionally, you can beat the simplicity.

Another filter I am starting to work in total (not open) credit lines > 14. Of course you take a hit on the number of loans but you can usually eek out an extra 1%.

Again, you need to weigh your filter criteria against how much money and how fast you want to invest. Great post.

Reply

Peter Renton December 19, 2012 at 5:53 pm

Thanks for sharing Michael. That is one thing that should be emphasized. If your filters only bring up 100 loans from all of Lending Club’s history then not only will there be few loans to choose from your sample size is so small that your future returns could be very volatile. I always like to see at least 500 and preferably 1,000 loans in the entire history for any filtering criteria.

The other point I should make is that Lending Club’s underwriting is a moving target. I have seen a lot of changes this year, so any future forecasting should keep that in mind. But if you can find a group of loans that consistently outperforms the average there is a good change it will continue to do so.

Reply

Jigglebilly February 9, 2013 at 9:00 pm

Good changes, or bad changes?

Reply

Bilgefisher December 19, 2012 at 1:45 pm

The difficulty I run into is availability of loans that give me the highest returns.

Peter, didn’t you review a statistic website awhile back that is geared towards finding the highest return possible?

Jason

Reply

Peter Renton December 19, 2012 at 5:59 pm

Hi Jason, Nice to hear from you again. I think the site you are thinking of is Interest Radar. They have a strategy ship that shows several ways to get a high return: http://www.interestradar.com/strategyshop

Reply

Dan B December 19, 2012 at 6:57 pm

Good job Peter.
Another few issues to keep in mind are that none of the return numbers cited are the “final numbers” yet. I say this without having to look at the breakdown simply because I know that LC originated 2.5 times more loans in 2010 than in 2009 & so I’d bet that the vast majority of the 1069 sample from ’09-’10 originated in 2010. So most of them still have several to many months to go. Also keep in mind that 5 year loans were introduced in May 2010 & in short order became over 50% of all loans during the 2nd half of 2010. So with these ’09-’10 loans (many of which were incidentally D-G rated), you still have 2 1/2 to 3 years left before they’re paid off.

5 year loans have since been scaled back, as we know, but they were super popular back then & into most of 2011 as well. I’m just saying……………………

Reply

Peter Renton December 20, 2012 at 10:47 am

Very true. I didn’t distinguish between 3 and 5-year loans in this analysis, and these ROI number certainly are not final. One should expect these numbers going forward but I think it is fair to say that you have a good chance of beating the averages regardless.

Reply

Zach December 19, 2012 at 7:02 pm

While I’m now a BIG supporter of Interest Radar and continually impressed by the work that the site owner has put in (and great results I see), I think Jason may be thinking of p2panalytics.com too?

That was the first website that came to mind.

Reply

Matt December 19, 2012 at 9:42 pm

I have found certain occupations and income above 50k to be great search criteria on Prosper. Look at my post on the forums regarding returns based upon occupation. It seems like their are numerous strategies for securing a high return. Peter, have the returns for repeat borrowers really fallen that dramatically?

Reply

Peter Renton December 20, 2012 at 10:49 am

Matt, Yes. I will have a post on repeat borrowers at Prosper early in the new year. Total returns have reduced substantially. There are certainly other filters like income and occupation that can be used very successfully to achieve an above average return.

Reply

Bilgefisher December 20, 2012 at 11:38 am

I was wondering about this. My default rate on return borrowers went up drastically in the last 6 months.

Reply

Anil @ PeerCube December 20, 2012 at 12:34 am

I interpret the Lending Club chart of return differently. The key information about that chart is in the footnotes.

The returns are dollar-weighted averages, i.e. higher loan amount and higher interest rate loans will have higher weight in the return of the grade. Most retail investor invest equal amount in each loan and not a fixed percentage of loan amount.

Average return for each grade is misleading. It is dollar-average of all loans issued in a grade. No investors lend to all loans. The calculations include the recently issued loans (3+ months). With the much higher volume in recent years, the average returns and upper bound on returns shown are inflated because of it.

Top 50% and Bottom 50% portfolio are created by selecting the best performing loans and worst performing loans respectively. The top 50% portfolio will most likely have recently issued loans and bottom 50% portfolio will most likely have oldest loans. Basically, what it tells that if your most loans are recent, your return will be in upper half and if you have an aged portfolio, your return will be in lower half.

But this is an important visual as it shows the spread/uncertainty of returns in each grade, most probably the only valuable insight from the chart. If you want better certainty in return (stable monthly cash flow), you are better off investing in high quality loans. After A and B grade loans, other grade loans could potentially generate net return lower than A and B grade loans, and a very high variance in returns.

A better Lending Club chart is the one that shows the expected range of return based on the number of notes in a portfolio. Lending Club can do better by running simulations and then showing the spread of returns for each grade and probability of achieving those returns (a good example is Firecalc simulation for retirement calculations).

Reply

Peter Renton December 20, 2012 at 10:58 am

Anil, I agree that taking the numbers in the chart as pure average is misleading and you are quite right to point out that these are dollar weighted averages. But my experience has been, with a well diversified and aged portfolio, that the my returns have followed a similar trend. That is, the lower credit grades (D,E,F,G) have outperformed the higher grades.

Reply

Dan B December 20, 2012 at 4:40 pm

Of course that is certainly to be expected (otherwise no one would invest in the lower credit grades) as long as one understands that the probable “out performance” comes with substantially higher risk & volatility,………………..as you pointed out.

Reply

Tim December 20, 2012 at 4:42 am

Peter, I have been investing in Lending Club since last May and recently took the 1k bonus offer at Prosper. The lending club account has done very well (13.22%) so far. It is too soon to know what to expect from the Prosper account. You mention the fiscal cliff. That is a little worrysome, but I am more concerned about the working capital at Prosper early next year. Do you have any thoughts or updates on that?

Reply

Peter Renton December 20, 2012 at 11:00 am

Tim, I have had many conversations with Prosper management about working capital and while no one will comment for me on the record, I expect we will see another funding round some time early next year.

Reply

RJL December 20, 2012 at 7:19 am

Interesting post. I think there are many ways to splice the data and come up with simple rules of thumb. However, it is quite difficult to rule out factors that may be correlative (with the appearance of significance) but in reality have no bearing on returns. I wonder if # of lines of credit falls in that category but I won’t claim to be sure.

In my experience, I still find much success in a strategy focusing on repeat Prosper borrowers with at least 6 on-time payments and no previous late payments, and income over $50K. A filter along those lines shows decent returns which is all well and good, but as Dan and Anil pointed out, the filter alone will never tell the whole story and should be taken a little suspiciously. Of near equal importance, the criteria passes the smell test of making sense (good income, proven experience with P2P lending). To me, the fact that Prosper provides data on repeat borrowers is a HUGE mark in its favor over Lending Club.

Anyway, too much detail but the bottom line for me is that any criteria must do well both when tested by imperfect data (the filter) and by imperfect reasoning (common sense). After that it’s trial and error.

Reply

RJL December 20, 2012 at 11:00 am

One more thing about the repeat borrowers on Prosper. I noticed that during the summer there was a promotion by Prosper for repeat borrowers to have one month paid. There was a spike in repeaters requesting loans in small amounts (around $4,000) at high rates (above 20%). Looking at my own portfolio, several of these loans are late or charged off. I wonder if this specific promotion is skewing the numbers as a one-time event. Just a thought.

Reply

Peter Renton December 20, 2012 at 11:06 am

RJL, Good comments. I still invest in repeat borrowers at Prosper but I have noticed a spike in defaults as well recently. It is an interesting point, though. Prosper does run promotions to repeat borrowers from time to time and it would be an interesting thing to measure if these resulted in higher defaults. My suspicion is that it does.

Reply

Peerlend December 20, 2012 at 10:50 pm

I hand select, so, I quite often go back and look at the credit details of the borrower’s previous loan and compare to their current details. Occasionally see repeats where their numbers suggest that they used first loan to pay off CC’s – ran CC’s up again – returning for a second loan. Needless to say I do not invest in these. But, I’d love to have an easier way to compare previous credit details to current credit details; as is, requires clicking on the user’s name, then scrolling down to past loans, clicking on it, often while using a new window to tab back and forth (if there is a lot to remember)…

Reply

Scott Stewart December 20, 2012 at 4:28 pm

Peter I’m with you. I’ve been slowly trying to both extend maturities and lower the credit quality of my prosper holding by selling my AA and A notes and moving into the HR notes. I’m currently avg. approx. 13% return avg for the last three years.

Reply

Tom December 22, 2012 at 11:13 pm

Hmmm…..why so hard on inquires? I would think anything 5 and over would be considered excessive. I used to work in new accounts for a credit card bank for 3 years. We would turn down customers for excessive inquiries if they had over 5 hard inquiries in the last 6 months, or maybe it was a year I can’t remember I worked there in the early 90s.

Reply

Tom December 22, 2012 at 11:16 pm

Correction,
I meant 5 or more inquiries not over 5.

Reply

Peter Renton December 24, 2012 at 3:36 pm

Tom, Check out this post from last year:
http://www.lendacademy.com/why-number-of-inquiries-is-my-favorite-p2p-lending-filter/

Even though this data is almost 18 months old I believe the same trend still applies. Lower inquiries = higher ROI.

Reply

Tom December 29, 2012 at 9:38 pm

Good point, I agree that there is definitely a correlation between the number of inquiries and the default rate. However, I think the real issue here is that Proper and I assume Lending club (I am only signed up with prosper) are missing an important filter; which is number of recently opened accounts opened in the last 6/12 months. This is a reason banks and credit cards turn potential borrowers down. As you state in your article,” If someone is out shopping for credit in several places and then they come to Lending Club or Prosper looking for a loan then they are a higher risk borrower. They may have some serious financial problems if they are shopping for a lot of credit.” Even worse they are trying to run up a bunch of debts before they go chapter 7.
I would limit my inquiries to 0, but I believe I would be trading off yield as they are probably getting hit on their ratings for the inquiries. I am more concerned about other things such as insufficient credit (less than three open accounts), and if they are delinquent now or have recent public records. For now I am going to limit my inquiries to 4 and below, but I may adjust this if I see a trend in the notes I hold.

Happy New Year!
Tom

Reply

Anil @ PeerCube December 29, 2012 at 11:11 pm

Lending Club provides several parameters that provide information related to “number of recently opened accounts opened in last 6/12 months.”

Accounts Opened in Past 24 months
Months since most recent bankcard account opened
Months since oldest bank installment account opened

Reply

William December 26, 2012 at 11:18 pm

I was just messing around on IR and found a that these three filters generate above average returns:
Amount Requested: 28-35k
Employment Length: 10+ yrs
Monthly Income: 8-20K+

# of Loans: 999
IRR: 12.6% – 14.8%

IR

NS

Reply

Peter Renton December 27, 2012 at 6:58 am

William, There are certainly many ways to take three filters and create a portfolio that will produce outstanding returns and your filters here meet that criteria. Monthly income is one of my favorite filters, one I use for my most of my investing, but I didn’t consider it for this post because it is not available at LendingClub.com as a filter. You have to download the BrowseNotes spreadsheet or use of one of the third party stats sites to invest in order to filter notes by income. But thanks for sharing.

Reply

Anil @ PeerCube December 27, 2012 at 10:20 am

And, you just demonstrated the dangers of blindly plugging the numbers and getting the results. This is the reason why I am proponent of understanding the underlying data to determine whether results are valid or not. No website or software package can substitute for human interpretation.

If you had reviewed the underlying data, you would have noticed that lending club didn’t start issuing loans above 25k until I believe 2011. That is most of these loans are not seasoned enough.

Majority of 25k+ loans are lower quality, I.e. higher interest rate. Also most such loans are 5 year loans. Once again, not seasoned enough thus lower default rate. Combined with higher interest rate, data will show higher return.

May be I should write a blog post.about it. :-)

Reply

William December 28, 2012 at 9:25 pm

Yea very true. I don’t uses that as one of my filters as I said I was just messing around. You should definitely make a blog post about it though.
Would it be possible for one of the stats sites to include an average age or something like that, so when a filter is applied it would tell you how old the notes are? It would make it easier for people to catch something like that.

Reply

Michael December 28, 2012 at 9:43 pm

William, I was watching this thread. This is easy to add. I just added it.

Take a look at http://www.nickelsteamroller.com/lendingclub_return

I added the average age in days to the result grid. This is based on the average issued date for grade population and also the input parameters if you specify any.

Reply

William December 29, 2012 at 9:45 am

Thanks a lot Michael this is really helpful!

Jigglebilly February 9, 2013 at 9:11 pm

I feel that the average age in days column is hard for less sophisticated investors to interpret. Surely, there must be a noob-friendlier way of showing the significance of that column?

Anil @ PeerCube December 29, 2012 at 2:46 am

William, thanks for the suggestion on writing a blog post. I might write one sooner than later.

Unfortunately, i dont believe average age would have highlighted this problem. I think what may have helped is to view the performance based on loan issued year.

Reply

William December 29, 2012 at 11:24 am

I think it does help, while it doesn’t solve the problem completely it does through up a red flag. Now if you click the NR link, you would see that the average age of the loans are not even 8 months old compared to all loans which have an average age of 14.5 mouths.
Yes I understand that LC is growing fast and the average loan age is getting pulled down by that. That is why this is not a solution merely a warning.

Jigglebilly February 9, 2013 at 9:08 pm

Holy crap Anil, that is an excellent point. Wow, thank you very much for that.

Reply

chasingbread December 30, 2012 at 11:44 am

Peter,

Nice post. I use to manually invest in all notes and I simply got tired. Are you using this filter?

Reply

SJM January 6, 2013 at 10:37 am

I am confused by the LC strategy. Your result shows 9.22% and 12.33% returns depending upon the period. However, if I use LC’s data and calculate a simple average of the D, E, F, G loans, I get a return of 12.57%. Does that mean that the second and third criteria are actually lowering the return?

Reply

Peter Renton January 7, 2013 at 10:37 am

SJM, I am using the return numbers from NickelSteamroller.com. This uses a formula where it discounts late loans. What that means is that if a loan is, say 16-30 days late there will be a loss factor of 23% used for that loan. In other words, you can expect your return to be 23% lower than on a similar loan that is current.

So, if you just use LC’s data and take all non-defaulted loans at full value then you will get a higher number, but that number will be less accurate. I hope that is clear.

Reply

Dan B January 7, 2013 at 4:17 pm

Am I understanding this incorrectly or are you saying that a 16-30 day late loan at LC is discounted in the calculations by only 23%? That is rather optimistic, don’t you think?

Reply

michael January 7, 2013 at 4:33 pm

The loss rates are taken from:

https://www.lendingclub.com/info/statistics-performance.action

Complete loan performance history is not something investors can download, so once a loan goes through all the stages of defaulting, that information in lost. It’s not possible to see if a loan was late and then went current. To my knowledge there is no deterministic way to produce loss rates from their exports.

I do allow you to change your own loss rates, as well as use LendStats’ rates.

http://www.nickelsteamroller.com/lendingclub_return

Is 23% optimistic? Probably. I was playing with the idea of allowing investors to vote on loss rates and produce crowd sourced loss rates instead of trying to guess. Perhaps since Peter’s blog has more exposure he could run an on going poll of what investors feel loss rates are. This could be a nice addition to his forum discussions as well.

Reply

Anil @ PeerCube January 7, 2013 at 5:22 pm

Loss Rate is a function of loan age. Any fixed guesstimate for loss rate is bound to be wrong. I don’t believe in any ROI numbers unless ROI is for a portfolio of matured loans, i.e. all loans are either fully paid or defaulted/charged off. That is what I include on PeerCube.

Reply

Dan B January 7, 2013 at 5:43 pm

Of course……………..but it is not so much a question of being absolutely accurate. There can be guesstimates that are within the ballpark & guesstimates that are in another county. No offense to Nickelsteamroller but I am saying that 23% seems almost like the latter. At the very minimum I’d say 40%…………….but 50% is imo more realistic.

Reply

Dan B January 7, 2013 at 5:45 pm

And once you get to the 31-120 late……………..then I”d say 70%+

Michael January 7, 2013 at 5:51 pm

No offense taken. I’m simply using LC’s published numbers. I feel as though I should not make a guess without empirical data to back it up. If people asked me “how did you get your loss factors” it’s probably more acceptable to say from LC than to say that’s what my gut tells me. I hope you understand why I use the numbers I do. I do allow people to use their own presets and I have even pre-programmed LendStats’ loss factors into my application for ease of use.

Peter Renton January 8, 2013 at 7:39 am

One thing I would caution about using your gut on numbers like this is that your gut is probably going to overestimate the percentage. I would agree that the number of loans that go 16-30 days late then default would be closer to 50% than 23%. But the amount lost is going to be far less than the absolute number that go late because on most loans some payments will have been made. Just something to keep in mind.

And I think Michael made the right decision to go with official numbers from LC but then allow the users to tweak the loss rates themselves. I just wish LC would update these numbers on a regular basis so we can see how they are trending over time. I have told them about it numerous times but no action yet…

Jigglebilly February 9, 2013 at 9:19 pm

Peter, I saved the loss rate data from Oct 2010, you can see here:

http://americaninbusan.com/?attachment_id=1677

The recovery rate of LC’s flaccid collections agency has improved insignificantly about a year later:

http://americaninbusan.com/?attachment_id=1676

I don’t know why they don’t have a more recent update. Perhaps they are hiding something important.

Reply

Peter Renton February 11, 2013 at 11:34 am

Jigglebilly, Both of those charts are now woefully out of date. I have been pushing with LC to provide some more transparency in this area but so far I have not had much luck. I did get to chat with their head of collections a few months ago and I think they are putting more focus into collections now than in the past:
http://www.lendacademy.com/the-collection-practices-at-lending-club/

Reply

John February 27, 2013 at 7:20 pm

Thanks for the simple strategy. I just signed up for LC and have to invest $5000 to get the $200 promotion bonus within a few weeks IIRC. Using similar criteria as you, the amount of loans are very limited. My question is if you were in my situation, what’s the most you’d put into each loan? Thanks!

Reply

Peter Renton February 28, 2013 at 7:39 am

With $5,000 invested I would stick with $25 per loan. There are fewer D-G loans these days so if you are in a hurry I would add C-grade loans to the mix. This is exactly what I am recommending to my father in law who just added a large sum to LC.

Reply

Blake Salisbury April 4, 2013 at 10:33 pm

Hi I am new to Prosper and this is the route I must go because Lending Club does not loan in Oregon. I am dropping 5k in and applying the filter above and only 16 loans come up for a total of $400 ($25 a loan). Is this right? Do I just set it to keep auto investing on this formula until the remaining 5k is invested? I was hoping to get all money working at once but perhaps this is not possible.

Reply

Peter Renton April 5, 2013 at 3:21 pm

That sounds about right. I think it is wise to invest slowly and not put your entire investment to work in one day. Keep in mind that new loans are coming on the platform every day. So you can either setup an Automated Quick Invest where it will invest in loans for you or you can login 2-3 times a week and put your money to work. Regardless, I would be surprised if you were not invested in 200 loans within a month or so.

Reply

Leave a Comment

Notify me of followup comments via e-mail. You can also subscribe without commenting.

{ 1 trackback }

Previous post:

Next post:

Real Time Analytics