My Bondora Second Market Strategy

Bondora shows very good transparency and lets you download the entire loan dataset which is fantastic. This dataset includes all loans that have ever been up for investment and is the perfect playground for the number’s nerd.

Yes, statistics are a bit boring so let me do this work for you. What we want to do is to optimize our profits by reducing the risk of investing into loans that will eventually default. The loan dataset gives us the chance to look at the performance of thousands of loans over the past years, so let’s get to it. If you don’t care for the numbers and only want to see the conclusion, just jump to the end of the text.


For my analysis, I looked at defaulted loans from 2016 to 2018. Bondora counts a loan as defaulted when a scheduled payment is more than 60 days late. Hence, the last payment was made 60 days before the default date.

At first, let’s see how long it takes loans to become default:

Bondora: number of defaulted loans vs. loan duration
Bondora: number of defaulted loans vs. loan duration

What we see here is the period in months before a loan becomes default and the number of loans that failed in the respective month after the loans were listed on Bondora. It’s obvious that the vast majority of the loans fail relatively early on within the first few months. Let’s take a closer look and turn this graph into a default probability plot which is much more suitable for our purpose:

Bondora: loan default probability vs. loan duration
Bondora: loan default probability vs. loan duration

Here we see much more clearly that most of the loans fail early on. For example, after 7 months, about 60% of the loans became default. This means that we can dramatically decrease our risk of buying a loan that defaults at some point by not buying “new” loans that have just been listed on Bondora.

There’s a gap of two months (60 days) between the last payment and the loan being labeled as default. Since we want to invest in active (current) loans with punctual payments we need to correct the time axis by those two months in order to better judge when we should buy a loan.

Bondora: loan default probability vs. active loan period
Bondora: loan default probability vs. active loan period

Following our example above, let’s say we only buy loans that have been active for at least 5 months. On average this will decrease our risk of buying loans which will become default by 60%. Pretty neat, right?

To refine this analysis a bit further, here’s the default probability for the different loan markets (Estonia, Spain and Finland):

Bondora: loan default probability vs. active loan period per country
Bondora: loan default probability vs. active loan period per country

The differences between these markets are quite significant. Spanish loans fail very early on and by buying loans with at least 5 months maturity you can sort out almost 80% of bad loans. For Estonia and Finland it would be better to buy loans with higher maturity to avoid buying too many bad loans.

2nd market strategy: picking the good loans

The selection of loans that I explained above can only be done on the Bondora Secondary Market. The portfolio managers for the First Market don’t allow for the required loan filtering and loans with sufficient maturity can’t be found on the First Market anyway.

Picking well performing loans doesn’t come for free. Basically all of these loans are sold on the Secondary Market with a premium or markup. This is compensated for by the better loan performance that you will get but I still wouldn’t pay more than 4% markup. It’s best to look at the effective XIRR which is listed for every loan on the secondary market. This number tells you the return of investment over the entire loan duration and is the best method to judge if a loan is worth buying or not.

Here’s what I propose to do:

Filter the loans on the secondary market based on the data I showed above. The screenshot shows how this could look like for loans from Estonia.

  • set the “Loan listed” date to before 5 months ago
  • Set loan status to current
  • Select the desired rating (I chose B, C, D)
  • Select the desired country (here Estonia)
  • Limit the markup (with up to 3-4% markup you’ll find a lot of available loans)
  • Set Has new loan schedule to No
  • Save the filter for easy re-use; don’t forget to adjust the Loan listed date when you use the filter again
Bondora Filter Settings Secondary Market
Bondora Filter Settings Secondary Market

You’ll get a long list of available loans. The XIRR value tells you about the overall profit if the loan gets fully repaid and includes already the markup you pay for the loan. You can further filter this list for a desired profit range by limiting the interest range in your filter settings.

Bondora filtered loans Secondary Market
Bondora filtered loans Secondary Market

Buy the loans you want by adding them to your shopping cart. For the sake of diversification, I’d recommend to avoid loans with a large principal (e.g. above 5€) and to invest in as many loans as possible. This is a manual process and cannot be automated, so I do this only about once a month when my account has filled up with repaid cash.

This strategy has been working very well for me: since I started using it my returns on Bondora have significantly increased compared to the portfolio that Bondora’s Portfolio Manager had invested in. It is not a convenient passive investment method like the Portfolio Managers but it pays off as you can see in the charts below. I think getting active once a month is a small price I pay for a much higher return on my investments.

My Bondora portfolio performance with and without Secondary Market strategy
My Bondora portfolio performance with and without Secondary Market strategy
Please follow and like us:


  1. Very interesting post. Do you put any restriction on the type of income/expense verification method?

    Best regards,

    1. Author

      Hi Yashar, good point. I actually used to consider income verification when I was investing in the first market but I didn’t do it for the secondary market. Would be easy enough to include this parameter in the statistical loan analysis though to see if these loans fail earlier/later/less/more. When I have a free minute I’ll give it a look. Thanks, Marian.

    2. Author

      Ok, I did a quick check and it turns out that it is better to take loans that don’t have income/expenses verified! Across all 3 countries (EE, ES, FI), loans with verified income/expenses have a failure rate of 37.5%. Loans without this verification fail to 28.7%.

      For Estonian loans this is less pronounced: unverified loans fail to 21.7% and income/expenses verified loans to 22.2%. But loans with only income verified (verification type 3) fail to 39.8% so these should be excluded if possible.

  2. Sorry of my late reply. I thought I would receive an email when there is a reply posted. Thanks for investigating this. This is an interesting point although I am not sure if the 9.2% difference in this case should be considered statistically significant. So far I have been only investing in Estonian verified (income and expense) loans. I will try a slightly different by ruling out only type 3 loans. Let see if it will make a meaningful difference.

    1. Author

      I added an option to subscribe to comments. Thanks for pointing this out!

      I do think that the 9.2% are statistically significant since the analysis is based on thousands of loans. But for Estonian loans it doesn’t matter anyways. I also decided to stay away from Spanish and Finnish loans. Takes too much effort to be successful there I figure.

      1. Hi Marian, awesome post – very useful

        Do you (or can you filter) if in the first 5 months till you bought them, all loans were paid on time / or had late payments? I think if all payments were made on time, it should add a little bit of security

  3. Hallo Roland,

    ich habe das mal für Rating D gerechnet:
    Von 100 Krediten bleiben nach deinen Ausfallraten also am Ende der Laufzeit übrig:

    EE: 100-> 72
    ES: 100-> 62
    FI: 100-> 50

    Wir steigen jetzt nach 5 Monaten ein. Nach der Grafik (ich lese ab, wieviel Prozent der AUsgefallenen Kredite dort sieht es so aus:

    Von den Estländern sind weniger Kredite (13) ausgefallen, aber es werden noch 15 Kredite ausfallen. Bei den Spaniern sind sehr viele Kredite (29) ausgefallen, es werden aber auch nur noch 9 Stück ausfallen. Bei den Finnen sind 25 Kredite ausgefallen und es werden noch einmal 25 Kredite ausfallen.

    EE: 100-> 87 -> 72 Noch 17 % Ausfälle
    ES: 100-> 71-> 62 Noch 13 % Ausfälle
    FI: 100-> 75 -> 50 Noch 33 % Ausfälle
    Damit schneiden die Finnen am schlechtesten ab, die Spanier sogar am Besten.

    Ich rechne an diesen Dingen schon seit Tagen rum und glaube es mittlerweile irgendwie durchdrungen zu haben. Jetzt braucht man nur noch eine sehr große Zahl an Krediten um (nach dem Gesetz der großen Zahlen) diese Werte auch im Mittel zu erreichen.

    Viele Grüße und weiterhin alles Gute für deinen Blog / Youtube. Finde ich toll!

    1. Author

      Hi Christian,

      genau, das stimmt so. Nur bei der Ausfallrate der estländischen D Kredite hast dudie Ausfallrate der E Kredite verwendet 😉

      Was die spanischen D Kredite angeht: davon gibt es eben fast keine weswegen man dann in der Praxis doch auf andere Risikoklassen ausweichen müsste.

      Leider ist die gezielte und automatisierte Investition am Zweitmarkt mittels API mittlerweile nicht mehr möglich. Und wiedu richtig sagst, diese ganze Auswertung gilt nur bei einer statistisch relevant großen Menge an Kreditanteilen.

      Danke für das feedback,


      Marian (nicht Roland)

  4. Warum sind Investitionen mittels Api auf dem Zweitmarkt nicht mehr möglich?
    Ich dachte genau dafür ist sie gedacht?

    LG Tim

    1. Author

      Der API Zugriff wurde auf einen pro Minute begrenzt. Theoretisch ist damit schon noch eine Investition via API möglich. Da aber nicht jeder Zugriff zu einem erfolgreichen Kauf führt ist der Durchsatz effektiv so niedrig, dass sich das nicht mehr lohnt.

Leave a Reply

Your email address will not be published. Required fields are marked *