Projections Using LendingClub's LendingMatch Slider

Although I’ve publicly wondered about non-retirement ways to invest in the past, I’ve now made a decision: LendingClub. Sure, my wife was skeptical at first (“You want to lend money to complete strangers?”), but LendingClub’s founder and CEO was able to persuade her.

It’s not the first time I’ve invested with LC. About a year ago I signed up for an account and received $25 as a promotion. I invested that money in one note as a test. The borrower had a good credit rating (B) and the interest rate was good (over 10%, if I remember correctly)—and, most importantly, all I had to lose was $25 of “free” money.

I forgot about this investment for a long time—almost a year. When I came back, the economy was in a recession and I had become used to seeing my investments show very large negative returns. The LC investment, on the other hand, had a net annualized return on investment of 10.75%. Needless to say, I didn’t need the video to convince me it could work as a serious investment mechanism. (Yes, I was willing to assume I didn’t just get lucky.)

Now what?

Now that I’m going to invest more money, though, I’m faced with a new challenge: How do I best distribute the funds? On the site, I have the option to browse the notes and invest in them individually, but there’s also a nifty tool called “LendingMatch.” It takes an investment amount as input and then allows you to use a slider to pick a “target average interest rate.” As you move the slider from low to high, the number of notes poor-credit-score loans increases, as well as a risk factor (that’s what I call it, at least). There is also a listing of the number of loans used, which increases from the lowest target rates, plateaus in the middle, and then decreases again at the highest target rates.

This risk factor piqued my interest. It was listed as x/1 – where x raged from .11 to .51 at the time I looked. I wasn’t sure of the deeper meaning (and there wasn’t a detailed explanation immediately available on the site), but I suspected it was meant to show the relationship between the listed rate of return and the actual projected rate of return (which includes defaults). If this was the case, I should be able to plot the values at each point on the slider to find the best combination.

Results

Assuming ProjectedRate = TargetRate * (1 - x), here is how things looked using an investment amount of $1000 at about 2pm CST on 7 May 2009:¹

The total variance isn’t very high, with the lowest projected return being 6.93% and the highest 8.66%. In fact, if you ignore the last four data points, all projected rates are within one percentage point.

It’s also worth noting all of the projected returns beat the socks off any Money Market, savings account, or CD of which I’m aware.

Interpretations

If you ignore the high end, its almost as if LendingClub was trying to keep the projected rate of return consistent. That dip at the end, though, throws a wrench in things. I wonder if it indicates an exponential relationship between credit score and default rates. In other words, those with excellent credit default on their loans at only slightly higher rates than those with good credit, but the horrible credit folks default at much higher rates than their poor credit neighbors.

Alternatively, maybe the most influential factor is the number of loans available. The most loans were available in the middle, while about half as many were available at the extremes.

A second run

I went back an hour later and the numbers had already changed (presumably because certain loans became fully funded and others were posted). Here’s how things looked this time around (using the same investment amount on the same day, just at 3pm CST):²

Once again, the highest projected rate of return (which was 8.66% in both cases, oddly enough) comes with choosing a target rate somewhere in the middle—12.55% last time and 12.73% this time.

I paid attention to the number of loans involved at each point this time, too. Clearly, there is an upward trend:

Loan distribution

One thing I didn’t expect was for there to be some poor credit loans involved in the “best” allocation of my funds. Consider the distribution of the 40 loans on the second run:

Working the percentages, this indicates one F loan, three E loans, and eight D loans – all loans I wouldn’t even have considered if I was doing my own browsing/investing on a loan-by-loan basis. Now that I think about it, though, it makes sense. You’re supposed to keep things “diversified,” right?

Closing

In the end, this could all be a bunch of hooey, as I’m not at all sure my approach is valid. It is interesting to try and get a glimpse into the inner workings of the site, though.

For more information on how LendingMatch works, I found these two articles on their blog:

Notes

¹ My raw data for the first run:

Target Rate	Risk	Projected Rate
8.92%	0.11	7.94%
9.16%	0.12	8.06%
9.39%	0.13	8.17%
9.63%	0.14	8.28%
9.89%	0.16	8.31%
10.10%	0.17	8.38%
10.32%	0.18	8.46%
10.55%	0.20	8.44%
10.80%	0.21	8.53%
11.04%	0.23	8.50%
11.24%	0.24	8.54%
11.50%	0.25	8.63%
11.77%	0.27	8.59%
11.98%	0.28	8.63%
12.14%	0.29	8.62%
12.55%	0.31	8.66%
12.83%	0.33	8.60%
12.99%	0.34	8.57%
13.11%	0.35	8.52%
13.40%	0.37	8.44%
13.65%	0.38	8.46%
13.88%	0.40	8.33%
14.13%	0.42	8.20%
14.35%	0.44	8.04%
14.59%	0.46	7.88%
14.82%	0.48	7.71%
15.05%	0.50	7.53%
15.27%	0.51	7.48%

² My raw data for the second run can be found as a Google Spreadsheet: “Projected Rates Using LendingMatch on LendingClub.” Accessed May 2009 at http://spreadsheets.google.com/pub?key=rCf8dMuAknVRynQi2c2FnuA.

This article was edited after publication by the author on 08 May 2009. View changes.