LendingClub Data |> Pandas |> Crossfilter.js, D3.js, DC.js
Online peer-to-peer lending platforms such as the LendingClub and Prosper have recently seen a great deal of growth.
Borrowers receive loans to refinance their credit cards, consolidate their debt or finance a home improvement at a better rate than their banks would provide. Lenders get a return on their investment that is typically much better than that gained from traditional Certificate of Deposit or Saving Accounts.
As an investor, you should build a portfolio by extending small loans to a large number of individuals in order to benefit from diversification.
Every month, you’ll collect interests and notional repayments from people you’ve lent to, and occasionally you’ll suffer losses from people who have defaulted on their payments. In typical cases, the losses will be less than the interest payments and you’ll be left with a good positive return on your investment.
When defining your investment strategy on the lending club platform you should essentially care about 4 things:
expected returni.e. how much money you hope to get out of your investment
riski.e. the uncertainty around these returns
maturityof your investment i.e. how long your money is committed for
liquidityof your investment i.e. how long it will take to deploy your cash on the platform, and to liquidate your positions if needed
The performance of your investment will depend on who you lend to and on the proportion of good payers and bad apples in your portfolio. The Lending Club helps you in making your selection by assigning a grade to each loan. The platform also allows you to filter which people you lend to according to some additional criteria based on location, credit and income data.
The most important criteria are:
People whose credit report shows a delinquency are heavily penalized by the Lending Club algorithm. A bit too much, it seems. That's why I like lending to them a lot. I limit myself to grades B,C,D and E. It gives me historical returns above 9%, with little risk as returns look stable through time. 413M worth of such loans have been issued on the platform in 2014, which represents about 11% of the total supply.
I also like lending to people who own their home, and whose credit report doesn't show any credit inquiry. They feel safe to me. Returns are also very consistent through time. Supply is good, 883M$ on the platform last year. For an extra ‘kick’ I remove the ‘A’ grade from the filter, selecting only grades B,C and D. It’s gets me up to 8.09% return.
In general, I don’t much like ‘A’ graded loans. Yes, they're safe, but they don't yield high returns. Their average return is 4.3%, whereas the average of the market is 6.68%. Therefore, although this is probably the only category that did not suffer excessively during the economic crisis, I find that the premium to pay for a grade 'A' is too high.
Watch for your average return (
expected return), consistency of returns through time (
risk), while making sure there is enough supply (
liquidity) on the platform to deploy your strategy.
The goal is to provide a simple interactive tool to explore and compare the historical performances of different investment strategies. When I first started to invest on the Lending Club platform, I was a bit frustrated by the lack of analytics that would help me make good investment decisions. This is a first attempt to address the problem.
Most investors deploy and re-invest their money continuously on the platform and therefore many own a portfolio with loans of different ‘vintages’. The ROI are computed to reflect this, as they are returns averaged across vintages.
Currently, when there are not enough loans for a given vintage in a given filter, the return analysis may not be statistically significant and may yield outliers. You may therefore find a couple of filters showing abnormal data points. This should be addressed by further normalizing the data.
Please also note than due to the low issuance volume in the early days of the platform, the returns computed for the pre-2010 period are much less reliable than the post-2010 returns.
For any questions, feel free to contact me at [email protected]