The fun part, for those of us who like databases, is that anyone can put all the data into a database and analyze it and graph it anyway they like. Right now only 18 months of data exist because Prosper has only been around for about 18 months. As time goes on, this data will become more detailed and much more accurate.
Theoretically, once the risk factors have been quantified for each item in a credit report, it should be possible, with a diversified portfolio, to calculate an approximate rate of return after accounting for expected defaults. Right now on the different websites you will see things like estimated ROI, Experian ROI, and EricCC ROI. Everyone uses different methods to estimate the expected return, and right now these can vary quite a bit. As more data becomes available these estimates should improve.
The current consensus based on analysis done to date is that on average HR loans return a negative rate of return after accounting for defaults. E's are slightly positive, and loans in the remaining credit grades average between 5-12% after defaults. Some people still go after the lower credit grades and try and cherry pick the better loans out of group with the hope that if they avoid too many defaults they will end up with a high rate of return.
I prefer to stick to higher credit grades with the knowledge that 5-12% is better than I can get in a CD or a savings account, and it is much more fun than buying a CD. It is a feel good investment knowing that there is a person on the other side of the loan that is being helped out by having access to needed money.
Analyzing Prosper data
Prosper has created a set of 3rd party tools that allow developers real time access to lending data on Prosper. This has allowed enthusiasts and academics to create all kinds of different ways of looking at the data. The most popular 3rd party site, LendingStats, allows you to view anyone's profile, get breakdowns of loans by state, credit grades, and group statistics. Other sites like Eric's Credit Community allow you to be notified when a lender places a bid so you can decide if you want to copy someone's movements, and lets you track "What if" profiles. Some websites have tried to quantify and graph exactly what the risk is associated with number of credit lines, number of inquiries, DTI, monthly income, and home ownership.