Prosper: Best Search Filters for Automated Quick Invest

(This post is for investors and lenders. If you need a loan or debt consolidation, check out my LendingClub vs. Prosper comparison for borrowers.)

As part of my new Beat-the-Market Experiment, I have dedicated $5,000 to Prosper. As a quick recap, Prosper.com securitizes person-to-person loans so that you can lend money to other people in $25 increments and earn interest. The idea is to replace banks and credit cards as the middlemen. Since their mid-2009 re-launch after SEC registration, there have been a full cycle of 3-year Prosper “2.0” loans fully maturing with an average net return of over 8% annualized. However, this is still unsecured lending which means no car or home as collateral, and thus there is a risk of loss (which can be mitigated by diversifying in multiple loans).

Prosper looks at the credit history of prospective borrowers and charges them an interest rate based on a Prosper Rating of AA, A, B, C, D, E, or HR (high risk). (The ratings are relative; the minimum credit score is 640.) Now, if Prosper’s grading system was perfect, life would be simple. The interest rate charged would be high enough to cover any defaults plus a little extra for the added heartburn. Ideally, after defaults and fees are accounted for, perhaps AA loans would earn 6%, C loans would earn 8%, and E loans would earn 10%.

However, things aren’t quite that neat. Prosper publicly shares all its loan information, and smart folks have made tools to analyze that data. Currently, the best place to go is Prosper Stats. If you take all the loans, we see that AA loans have a net return (after estimating losses from late loans and actual losses from defaults) of ~6%, C loans had a net return of ~11%, but E loans only returned ~9%. Hmm. Look further and you’ll see other small inconsistencies. For example, loans to people with 2 or less open credit lines actually have a measly 3% net return, while loans to folks with 18+ open credit lines open have net annualized returns of over 11%?!

As a result, many investors avoid investing in Prosper loans blindly and instead use specific search filters. Indeed, Prosper makes it easy with their “Automated Quick Invest” service which automatically invests in loans that satisfy your custom search rules. There’s no need to spend time every day looking for loans. So, what are some possible criteria?

• Prosper Rating. Actually, Prosper seems to have done a pretty good job overall with their post-2009 Rating system. The lowest net return is actually for the AA borrowers with their low interest rate and low default rates. Unlike in the past, it may be better to take on some risk here and skip the goody-goodies :).
• Number of recent credit inquiries. How many hard credit checks has the borrower had in the last 6 months? It appears that the best net returns occur with those with less than 2 inquiries. 2-5 is okay, as the higher interest rates seem to compensate, but above 5 and returns start to sink fast. As it turns out, I like to keep my own credit inquiries below 5 in the last 6 months.
• Previous borrowers. Has the borrower had a Prosper loan in the past? Repeat borrowers have more than a 2% better net return than first-time borrowers.
• Homeowner? Homeowners have a little more than 1% higher net return than non-homeowners. Perhaps a bit surprising for loans that started after 2009.
• Stated income. Borrowers that stated on their applications that their income was either zero or they were unemployed had horrible net returns. This risk was not well-compensated for; they need to be charged higher interest.
• Current and open credit lines. As noted above, for some reason borrowers with a 0-2 current and/or open credit lines have significantly higher default rates than those with more credit lines. I find this interesting, lots of possible explanations.
• Loan amount. Loans of over $20,000 have net returns that lag by over 2% annually. Bigger loans means bigger risk, again not adequately compensated for by the assigned higher interest rates.

As with all backtesting-based systems, you’ll have to ask yourself… will the past predict the future? In many specific cases, there may be so few loans that any differences aren’t really reliable. But I still feel that with a few simple and broad filters, you can still have a pool of slightly-above average loans. In case you’re wondering, Prosper loans are not anywhere near as efficient as the stock market as Prosper sets the prices and only individuals simply choose to buy or not buy. (There is also a secondary market, which I don’t actively participate in.)

These are just a few possible ways to filter your loans, feel free to share your own findings in the comments. Prosper advertises historical “seasoned” returns of 10.08%. I’ll be happy with 8% after all is said and done, and I’ll provide screenshots on a monthly basis. My own filters are based very closely on the factors noted above.