Great read. I don't know much about US regulatory environment but if you have to do something similar in India; I wonder the difficulties with data sharing from aggregators, different parties involved will just kill you.
Great article. As a Data science practitioner who is learning finance, this article is very helpful. As you mentioned, upstart's customers are from credit Karma. I would like to dig more, and see how is the relationship with this credit Karma, because I see the risk of credit Karma going to multiple platforms such as upstart in the future, which might damage the laters earnings in the future.
Thanks for the well written article. Besides expansion to other loan products, it would be interesting to know if they plan to expand geographically to other regions like Europe, Australia. That could reduce dependency on one bank and test the execution outside comfort zone of USA.
Thanks! Upstart has hinted in the past that they do want to be a global company and not limit themselves to just US but I think overseas expansion is a couple of years away.
Their immediate priority is to move from a single product company (personal loans) to a multi-product company (personal loans, auto loans, mortgage, small business loans).
Hey Tariq, Superb read as always. Thank you so very much for the insights you bring in. I just have a quick question, when upstart makes an entry into small business/mortgage lending wouldn’t there be already existing players who would then have an edge over upstart. We already have companies like drip capital in US or Reevoy in India which are into similar AI/ML based lending or factoring business. Wouldn’t they be having a larger market share considering them having an earlier headstart. Do you see any hindrance to upstarts growth?
The lending market is pretty big and Upstart acts more of a platform for other smaller and medium scale banks to quickly set up ML driven credit risk based lending, so in that regard I do not see very many large scale competitors for Upstart.
To give you an example, auto lending in US is a pain with plenty of paper work involved and the entire process takes a lot of time. Upstart its trying to replicate its success in automation of personal loans in Auto Financing as well. They brought Prodigy as they could immediately get a leg up in dealers that were already using Prodigy's software. Further Upstart is now working on to automate end to end auto financing part to the extent they can. These are added advantages and competitive benefits that Upstart enjoys over others.
Further Upstart has a large network of 100+ institutions that are always hungry to purchase loans and securitize them. These kind of capabilities are not always easy to build.
But Yeah, having read the upstart story. All big banks already have their own proprietary machine learning algorithms for Probability of Default and Loss Given Default. It is not as aggresive as Upstart’s given heavy regulations big banks face.
Upstart i think is bridging a big gap between small debtors who don’t qualify Big Bank’s criteria and small banks who don’t have in house access to sophisticated ML driven Data models.
Thats correct. Upstart's sweet spot for personal loans was to enable the debtors who werent able to qualify for a loan from big banks and enable small banks to find these credit worthy borrowers who were earlier left out due to inadequacies in the lending system.
bhai why cant the big banks start their own algos why to depend up on upstart? after a decade or two the old might get a lil redudant as well na? is my understanding correct bhai?
It's not easy to build a complicated ML model that can do this let alone a stack of differentiated ML models that Upstart uses
Building ML models on simple data is hard enough, Upstart's models leverages and consistently trains on relevant data that Upstart is generating daily
Data is the biggest treasure trove for Upstart followed by its ML learning algorithms
The big banks can certainly replicate what Upstart is doing but it's not going to be an easy journey and Upstart has a eight year headstart in this area.
One more reason its difficult for each bank to do this on their own is because they will only have their own customers data. This limits model accuracy. Upstart on other hand is dealing with multiple banks potential customers which will grow only in future making its model even more accurate over time.
As always, great article! I was wondering if you analysed "Affirm" while performing the research for this article. If yes, how would you rate it just in terms of business model?
I haven't explored Affirm in depth, had researched it a bit previously. They aren't a peer or competition to Upstart as Affirm is focused more on the micro lending and BNPL side of things. Affirm for example will never get into Mortgage Lending which Upstart views as its biggest market.
Great work by publishing the special situations dashboard. Request update it regularly, say monthly, so that investing public can benifit from it. I know it is too much of work, but i am sure u will take care of your baby (dashboard). Thanks
Insightful. Thanks Tariq for the great work.
Thank you
Awesome read only heard about upstart now having read this seems like a great fintech
Awesome blogpost Tariq.Thanks for sharing this wonderful analysis. Would love a similar in-depth analysis of Roblox.
Thank you, I will release it soon
Great read. I don't know much about US regulatory environment but if you have to do something similar in India; I wonder the difficulties with data sharing from aggregators, different parties involved will just kill you.
Great reading. Packed with lot of information which is not known to many. Excellent effort by you. Looking forward to next publications.
Thank you Biswajit! Glad you found it useful.
Great article. As a Data science practitioner who is learning finance, this article is very helpful. As you mentioned, upstart's customers are from credit Karma. I would like to dig more, and see how is the relationship with this credit Karma, because I see the risk of credit Karma going to multiple platforms such as upstart in the future, which might damage the laters earnings in the future.
As always, great article, very insightful.
Thank you Chandra
Good write up.
Thank you
Thanks for the well written article. Besides expansion to other loan products, it would be interesting to know if they plan to expand geographically to other regions like Europe, Australia. That could reduce dependency on one bank and test the execution outside comfort zone of USA.
Thanks! Upstart has hinted in the past that they do want to be a global company and not limit themselves to just US but I think overseas expansion is a couple of years away.
Their immediate priority is to move from a single product company (personal loans) to a multi-product company (personal loans, auto loans, mortgage, small business loans).
Hey Tariq, Superb read as always. Thank you so very much for the insights you bring in. I just have a quick question, when upstart makes an entry into small business/mortgage lending wouldn’t there be already existing players who would then have an edge over upstart. We already have companies like drip capital in US or Reevoy in India which are into similar AI/ML based lending or factoring business. Wouldn’t they be having a larger market share considering them having an earlier headstart. Do you see any hindrance to upstarts growth?
The lending market is pretty big and Upstart acts more of a platform for other smaller and medium scale banks to quickly set up ML driven credit risk based lending, so in that regard I do not see very many large scale competitors for Upstart.
To give you an example, auto lending in US is a pain with plenty of paper work involved and the entire process takes a lot of time. Upstart its trying to replicate its success in automation of personal loans in Auto Financing as well. They brought Prodigy as they could immediately get a leg up in dealers that were already using Prodigy's software. Further Upstart is now working on to automate end to end auto financing part to the extent they can. These are added advantages and competitive benefits that Upstart enjoys over others.
Further Upstart has a large network of 100+ institutions that are always hungry to purchase loans and securitize them. These kind of capabilities are not always easy to build.
But Yeah, having read the upstart story. All big banks already have their own proprietary machine learning algorithms for Probability of Default and Loss Given Default. It is not as aggresive as Upstart’s given heavy regulations big banks face.
Upstart i think is bridging a big gap between small debtors who don’t qualify Big Bank’s criteria and small banks who don’t have in house access to sophisticated ML driven Data models.
Thats correct. Upstart's sweet spot for personal loans was to enable the debtors who werent able to qualify for a loan from big banks and enable small banks to find these credit worthy borrowers who were earlier left out due to inadequacies in the lending system.
Good stuff. A very neat and practical application of ML which actually makes sense, unlike many of the so-called "AI powered" companies.
Thank you! Glad you found it useful
bhai why cant the big banks start their own algos why to depend up on upstart? after a decade or two the old might get a lil redudant as well na? is my understanding correct bhai?
Valid question
It's not easy to build a complicated ML model that can do this let alone a stack of differentiated ML models that Upstart uses
Building ML models on simple data is hard enough, Upstart's models leverages and consistently trains on relevant data that Upstart is generating daily
Data is the biggest treasure trove for Upstart followed by its ML learning algorithms
The big banks can certainly replicate what Upstart is doing but it's not going to be an easy journey and Upstart has a eight year headstart in this area.
One more reason its difficult for each bank to do this on their own is because they will only have their own customers data. This limits model accuracy. Upstart on other hand is dealing with multiple banks potential customers which will grow only in future making its model even more accurate over time.
Insightful, as always!
Thank you Shivam!
As always, great article! I was wondering if you analysed "Affirm" while performing the research for this article. If yes, how would you rate it just in terms of business model?
Thank you Karthik
I haven't explored Affirm in depth, had researched it a bit previously. They aren't a peer or competition to Upstart as Affirm is focused more on the micro lending and BNPL side of things. Affirm for example will never get into Mortgage Lending which Upstart views as its biggest market.
HI,what are the websites you use to research us companies?
Great work by publishing the special situations dashboard. Request update it regularly, say monthly, so that investing public can benifit from it. I know it is too much of work, but i am sure u will take care of your baby (dashboard). Thanks