Filtering Loans – Part 4: Splitting the Filter to Find More Loans

Part 1 – Introduction: Risk
Part 2 – Avoiding Defaults
Part 3 – Low-Grade Loans

Part 4 – Multiple Filters
Part 5 – From Tools to Platforms

Hopefully by now you understand the basics of finding borrowers who are less likely to default on their loans (part 2). Secondly, you have chosen a degree of risk for your account, perhaps choosing to invest in a few low-grade loans (part 3). Today we will try and combine these two tactics into a single filter, going on to split it into multiple filters. Additionally, even though today’s post focuses on Lending Club, all this information works just as well on Prosper’s site.

Before we start, it would be wise to declare a goal: today we will try and create a medium-risk filter set with a historical ROI of 12% (remembering both that this “historical” ROI is probably a bit inflated and that past performance is no guarantee of future returns).

Recreate Our Filter

Return to the NickelSteamroller Return Forecaster, and recreate the filter we made when we looked at lowering our defaults.

If you remember, our previous filter looked like this:

  • No Inquiries in the past 6 Months
  • No Public Records
  • 5+ years of employment
  • Monthly income $5000/month or more

Remove CA BorrowersLet’s modify it a bit more:

  • Exclude borrowers from California
  • Exclude loans purposed as ‘Business loans’ and ‘Other’

Our filter on the NSR Return Forecaster would look like this:

May 21 Filter

A return of 10.10% with a default rate of 1.37% is really great. Now let’s add a bit more risk by taking away the safest loans. Uncheck the A & B loan grade options, leaving only C-G grade loans. Hit filter and watch the ROI climb to 12.74%. The default rate also rose to 2.17% (the loss of those premium A & B grade loans means a higher default rate and volatility).

May 21 Filter adjusted

In the example above, we have successfully combined parts 2 and 3 of this filtering series, removing defaults and adding risk. Additionally, you can see that 5764 historical loans match this filter, which is a solid amount of loan history.

Splitting the Filter

Here is an important lesson: any time we section off large portions of the historical loan pool, we are almost certainly removing good loans with the bad. Multiple filters allow us to reenter previously sectioned off territory to discover good loans our filters left behind.

Let’s restate that our goal is a 12% historical ROI. Right now we have 5764 loans giving us 12.74%, but we know our single simple filter has cut off a bunch of good 12% loans. How can we rediscover them?

If we split this filter in two, we may be able to reach them, attaining an even wider historical profile that still gives us 12%. Using multiple filters often give us more loans to invest in today, an important option if there are few available loans on Lending Club’s website (a common problem) or if we are trying to invest in lots of loans per week.

Let’s begin. Split the filter down the middle using the Monthly Income attribute, making one filter $5000-$6999 in monthly income, and the other $7000+ in monthly income. You can use any attribute to split your filter, but I prefer more numerical options like a person’s monthly income, because you can adjust it to be evenly split.

New Filter #1 | Monthly income $5000-$6999:

May 21 Filter 1

New Filter #2 | Monthly income $7000+:

May 21 Filter 2

The benefit of having split this filter is that we can adjust both of these in different ways to meet our goal.

Adjusting Filter #1

Filter #1 is actually a bit below our goal of 12%, so let’s tighten it down to get that ROI up (remembering to stay above 2000 historical loans). Change the Employment Length filter to be only 6+ years of employment instead of 5+ years. If we hit filter, the ROI rises to 12.21% with 2792 historical loans.

May 21 Filter 1 adjusted

Adjusting Filter #2

Now let’s look at filter #2. Its historical ROI is 13.56%. We can loosen this filter so as to find more historical loans (and thus have more results when we filter the platforms today). Change the Employment length to include all employment (check all). Then recheck the box for ‘California’ and include ‘Other’ in Loan Purpose (leaving ‘Small Business’ unchecked). Hit filter:

May 21 Filter 2 adjusted

Our filter ROI lowers to 12.46%, but our number of historical loans rises to 5031.

Let’s take a step back for a second and admire our work. Whereas before we had a single filter with a 12.74% ROI and 5764 historical loans, now we have two filters that both have a ROI of 12% but return a combined 7823 historical loans. Using multiple filters, we found 2000 loans that we previously would have missed.

In summary, here are the two 12% ROI filters we developed in this post:Lending Club Filters from May 21

Feel free to right-click and save this graphic (remembering that past performance is no guarantee of future returns). Again, the benefit of having broadened our filter into two is that, when searching the Lending Club loan pool with both of these filters, we should have a wider selection of 12% ROI loans to choose from.

Reducing & Adding Risk

Do these filters have too much risk for you? Or perhaps, like me, you want to add more risk. You can change the loan grades to match your chosen level of risk. For instance, many people prefer more stability. If they select only A & B grade loans, the default rate drops from over 2% to around 1% – a much more stable loan history.

Filter #1 with only A & B grade loans:

May 21 Filter 1 ABs

Filter #2 with only A & B grade loans:

May 21 Filter 2 ABs

Considering these filters return a combined 15,000 historical loans, you could probably use them to invest large sums of money in the platform.

On the other hand, you could earn a much higher return if you focus on only E, F, and G grade loans like me. Just remember that this loan pool is the most volatile. It has higher variance, meaning there will be a wider variety of ROI between the different lenders who go that route.

In our final section, we focus on moving from these filters onto the actual websites themselves, investing in some filtered loans that have met our criteria.

Questions or comments? If you enjoyed this please Like or Tweet it below.

[image credit: Mike Smail “Two roads diverged in a non-yellow wood” CC-BY 2.0]


Profile pictures via Gravatar

  1. William says

    Simon, thanks for the great info. After checking out Nickel Steamroller, combined with some complaints I’ve noticed, investors using this historical return tracking tool should only look at “Issued Date End” that is at least one year ago. Otherwise, the loans are not ‘seasoned’ and have significantly greater risk of default. As of today, when the “Issued Date End” is set to May, 2012, the average total ROI drops from 7.94% to 6.89%. Since so many of these loans were made in the last year, including loans made less than one year ago will create an unrealistically high ‘historical’ ROI.

Leave A Reply