UPST - some math

The private investor at the very beginning of the call asked how we can think about Auto’s contributions to revenue in 2022 and 2023.

Sanjay responded that it will be meaningful but too early to start giving numbers. I would expect the next quarter we should get at least some kind of guesstimate for FY22.


I listened to the call and frankly, I fail to see what is so impressive about this company. I thought the answers to every question was vague and uninspiring. I don’t own this and am glad I don’t have seeing the after hours action. I may wait a few days and take a starter position if I am come to gripes with this is something with a real future. They are going to have massive competition breaking into auto loans. I also don’t believe their loans will prove to be any better in risk in the long run. Free money everywhere right now. Once a recession hits we will find out how good their models really work.


I don’t own this and am glad I don’t have seeing the after hours action
Even after dropping by 20% AH, it is still almost double the price from the last earning call. I swear I will jump in every stock that has such kind of return in such a short period, even if I know it will drop 20% a day in the future.

Free money everywhere right now. Once a recession hits we will find out how good their models really work
I never expect it is gonna work well in a recession. And I can tell right away that if a downturn happens, UPST will head down first. I think most people already knew this stock is not recession-proof. Its disappointing Q2-2020 is clear-as-day evidence. I also don’t expect their model will work as well in recession but I do expect it works “better” than the traditional FICO model. We are not in a recession, and as long as the thesis is still intact, we still invest.


What I wish they had shared with us was metrics on bank partners. It’s grow there that interests me; loan application counts are going to be driven by signing up bank partners… and we got no info on that front, that I have seen.


The fraud checking that needs to be done needs to happen irrespective of AI, automation, or human delivery. And there are pretty good solutions for doing that detection. Something that is automated, new or fast is going to get a lot of ‘attempts’ by do-badders trying to find a way to get fraudulent loans…this isn’t new, it just might be that the sheer volume of attempts and attacks took some additional controls on the part of upstart.

Also, separate note, 2 hours after earnings release, it turns out there was a typo in the earnings release data table:
CORRECTING and REPLACING Upstart Announces Third Quarter 2021 Results

SAN MATEO, Calif.–(BUSINESS WIRE)–November 09, 2021–
In the table titled RECONCILIATION OF GAAP TO NON-GAAP FINANCIAL MEASURES, in the column labled Three Months Ended September 30, 2021, Adjusted EBITDA should be $59,139 (instead of $29,139).

This is speculation on my part, but I would just note that there may be other types of automation out there that may have read that information and may have precipitated some of this after hours drop… tomorrow morning will tell.


* They said they had to remove fraudulent loan requests to get 23%; otherwise, the number would be 13.5%, significantly smaller than 24.4% from the last quarter. This number from previous

367,780 loans made in the quarter. The math says that the number of fraudulent requests went from “insignificant” to 1,125,000 in one quarter. 12,000 per day. This couldn’t be a serious attempt to defraud and walk into the sunset with bags of cash. Would be malicious attempt to clog the system, wouldn’t it?

Whatever, UPST weathered it.



Late in the thread but I feel compelled to point out comparison to Q3 '20 based on the S-1 they filed and summarize as, what is the panic about?

YoY Revenue growth for Q3: 228M / 65.3M 350%

YoY EPS growth for Q3:(normalized to 54.4 public float shares outstanding): 0.60 / .178 237%

Positive surprises: 7% on revenues / 90% on EPS (average of SA’s and Zack’s .33 / .31 #s respectively)

Am I missing something?

(And now there’s a neat 133% “oops” up notification in EBITDA?)



Just a lot of high expectations going into earnings and a big “miss” on expected Q3 revenue since everyone expected Q3 revenue to be +30% sequentially. When it came in “just” 17.8%, everyone panic sold. However, given guidance and their tendency to beat guidance by a decent amount, I think the guidance indicates revenue will be around $281 M if they have a 6% beat which is 23% sequentially. I think that’s pretty solid. I sold initially and decided to buy back my shares after reviewing the numbers a bit more.


Here is the EBIDTA correction mentioned above:…

Hopefully this helps.

Flying Circus,

Yes, I agree. A good example of how perceptions and reality are always intertwined.

This is a good example, though many others can be substituted:

“everyone expected Q3 revenue to be +30% sequentially”

Let’s be very clear that “everyone” seems to be one part of this board. Not analysts. Not everybody here either. Wording it this way is convenient but it won’t help preventing the next case of exuberant expectations:

–based on Q2 2021 vs Q2 2020, which was a unique comparison (same as with LSPD),
–based on Google Searches (which is not research as I hope everyone here knows very well),
–based on hope that certain numbers are realistic w/o considering the lack of “first party data” for said projections.
–allocations based on unrealistic expectations.

Over-reaction one way naturally leads to over-reaction the other way.

Personally, as of yesterday I had 10% in AFRM and 5% in UPST so I need to see what AFRM does before I can do anything about UPST. I also bought 5% of MNDY on, well, Monday of course, and I quite like the results there. Added to U after hours yesterday, which is proving a lucky move. I had 1% experimental position in AMPL, which will remain as is. FUBO remains as is (was 9%). For me, AFRM and FUBO need to rock every quarter.

UPST, on the other hand, I see as long-term holding and nothing yesterday changed that.