Prosper.com: Convincing My Wife, Part 2

...she ain't convinced yet.

In my continued efforts to convince my wife that Prosper.com is a good investment, I'll analyze another aspect of the website today. Today I'll study what makes the successful lenders successful, what makes the average lenders average, and what makes the biggest losers, well, the biggest losers. I'll be moving my analysis platform to a fabulous website that focuses solely on Prosper.com lender and loan data, EricsCC.com.

To get things moving along quickly, consider the following graph that shows all lenders' rates of return on a seemingly normal distribution curve (please click any graphic to enlarge it):
As you can see, the majority of lenders are making money, and a significant majority are also earning a higher rate of return than they would earn in a traditional savings account. However, of all the non-average lenders, there are more that are doing exceptionally poor than doing exceptionally well. This indicates that if you do not follow a reasonable, disciplined investment strategy, you are more likely to lose at a high rate vs gain at a high rate. I guess the same could be said about the stock market. Essentially, it's easier to make mistakes than it is to get lucky.

Do you ever watch that show called The Biggest Loser on NBC? Well meet the biggest loser on Prosper.com: scoobydoo. Here is a graphical representation of his investments:
As Antonio from the Merchant of Venice would say, His "ventures are in one bottom trusted." This guy has invested a lot of money into Prosper.com and has given several large loans to people with C-grade credit. If one or two of those loans defaults, his ship will have sunk.

Let's look at another big loser's profile. How about jasonpeery:
Here's another guy that has a poor, lazy investment strategy. He has invested over $50,000 in Prosper.com listings and has scores of late payments and defaults. This guy has made several individual loans over $1,000, including one that is in default for $11,000! Why in the hell would you EVER loan $11,000 to a person with high-risk credit? And without even asking them a question! I sure hope that jasonpeery is better at personal finance than he is at determining to whom he should lend his money. As Neil Boortz would say, I bet that this guy has a lot of rent-to-own furniture in his house. My guess is that this guy's grandmother died recently and left him a bunch of money. No one that worked for $11,000 and saved it would ever be that careless in giving it to a single high-risk stranger.

One thing to remember about Prosper.com's fee structure is that all individual loan fees are passed along to the borrower except for a 1% loan servicing fee which is paid by the lender. This means that, statistically speaking, there is no reason to invest more than $50 in ANY candidate. Period. If I lend $500 to one person or $50 to ten people, I will pay the same loan servicing fee. And though I may save a little time by investing more money in lower-risk candidates, it's just plain silly to not diversify to the max with sub-prime borrowers.

OK, so let's look at someone with an average return. Consider the portfolio of helpishere777:
Ahh, this is refreshing. This user is right in the middle. He is earning about 11% interest, which takes into account the probability of his late payments going into default. He has invested the same $50,000 that our last big loser had invested, but in a completely different way. Look at the nice even relationship between all of the blue and green lines. Do you know why they're all equal? Because he invested the same $50 into every single loan. He understands that in order to mitigate his risk, he needs to diversify -- especially if he can do it at no additional cost!

Now let's look at the best lender. I'm not going to evaluate the person earning the highest return on his money. Currently that person is DrakeCO, who is earning about 33.6% interest. However, the average length of his loans is less than one month and most of his loans have been large amounts (max of $1,500) to high risk borrowers. Because of the youth of his loans and the nature of his strategy, he is bound to fail. Instead, I'm going to look at someone earning about 20% return with a reasonably large average loan period (if it's not old, the borrowers don't have time to be late!) and a significant amount of money. It looks to me like the golden child of Prosper.com is brother_tam. Here is his portfolio:

brother_tam is obviously smart and probably a little lucky. He has invested a little more than $10,000 in Prosper.com, mostly in $50 increments. Of his 224 loans, he has given more than $50 only 13 times, probably just to spice up his account. As a lender that understands the need to diversify. He is aware that he can invest in lower-credit borrowers because of his discipline. But he doesn't invest in only low-credit borrowers. He has a nice normal distribution of his loans that has a mean slightly on the low-credit side.

To be a successful lender on Prosper.com, you need to stick with a disciplined strategy that is formulated around the values of diversification and a normally distributed loan strategy. When choosing which loans to bid on, consider your current portfolio and establish a quota. "Right now, 75% of my loans are to high-risk borrowers. I should invest in some low-risk borrowers."

Remember: there is no penalty for investing the minimum amount in a person. And with more than 2,300 active listings, you shouldn't run out of people to lend to.

If she's still not convinced, I'll have to write more tomorrow.

Why Social Networking is Killing the Traditional Survey

TechCrunch announced today that LinkedIn is going to begin its own ad network. Non-personally-identifiable information about users' age, income, education, etc will be placed into cookies that will be shared with partnering sites to help target certain ads based on a user's profile.

This is quite similar to the idea that I discussed last month about advertising based on financial profiles. The premise of my idea is to develop an exceptionally "sticky" personal finance website that would make assumptions about its users based on their financial transactions and then advertise appropriately.

Networking sites are starting to understand that their users are spending an incredible amount of time online, entering terabytes of their personal information in hopes of developing a new life-changing business or social relationship. The more useful features that these sites develop, the more information they'll be able to obtain about their users. For instance, if Facebook were to place a mapping feature on their users' photos albums (like Picasa has), they could see where their users travel and sell that information to local advertisers.

Back in the 90s, companies would try to obtain this kind of information by sending its customers index cards that asked questions about income, address, etc. Retail stored would ask for a zip code or area code to find where its customers were coming from.

Today, social networking sites are becoming Survey Central, and their users may not even realize it. By sites developing new features that are perceived as useful/fun/cool for users, they're opening a door into yet another dimension of personal information. By providing users with a useful or fun product, sites can collect the honest, accurate information that traditional survey adminstrators couldn't dream of. Take LinkedIn for example. As many join the site in hopes of being discovered by the Donald or at least scoring a lucrative business deal, their descriptions of their work experience are likely to be pretty accurate. Just as most people place reasonably accurate and verifiable information on their resumes, the professionals on LinkedIn will enter similarly accurate information into their profiles.

Social Networking sites should work with their key advertisers to figure out what information would be the most valuable to them and develop "back-door" applications and features to try and extract that information. Bars in New York City may be interested in advertising to Facebook users that are between 21 and 28, live in NYC, and have listed their interests to be "partyin" "drinkin" "gonig out" or "chillin." A government contractor looking to fill a handful of positions might ask LinkedIn to advertise the positions to users that log in frequently, are interested in job offers (a metric tracked on LinkedIn), and have at least five years of experience working for the federal government.

Because of user-created content on social networking sites, the way that companies collect data is doing a complete 180. Instead of asking 20 specific questions in a survey and deducing a conclusion about the participant, companies can inductively analyze a collection of user-provided data and target groups and demographics based on a predetermined strategy. Because of the nature of social networking sites, potential customers are already answering unasked questions. Now all advertisers need to do are figure out how to ask the questions and fill in the answers accordingly.