OT - Upstart (UPST)

Stocks A to Z
1

Fin-tech Upstart (UPST) has been tanking hard the last few days with no news and it reached a level where I decided to take a position. The company, founded by ex-Google leaders, brings an AI, machine learning approach to qualifying applicants for loans, and they have web-facing software making application easy and quick, branded under their own name and also available to loan partners to use, branded under their own names. UPST collects fees from loan partners for the approved applicants.

The company has had remarkable growth since they came public, and stock soared sky high, but has since tumbled back to earth even as growth continues and the company earns good money. The fear seems to be that rising interest rates and potential recession will interfere with their business, possibly mess up their ratings, though such fears donā€™t seem to have much basis in reality as far as I can tell. It mainly seems to be strong downward price momentum is shaking out more and more holders and feeding the fear in a positive feedback circle.

UPST was a darling of the Saul board, but a lot of them bailed out as the price dropped and dropped and dropped some more. There are still a lot over there who like it and they rebut the nay-sayers pretty effortlessly, IMO.

UPST initially entered business by qualifying applicants for unsecured personal loans, which has gone great, and now theyā€™re moving into the auto loan arena in volume. Next up are small business loans and small dollar loans before eventually attacking the biggie: home loans, which will probably be a couple or few years off.

I bought a position today with average price ~$74/share, which is PE of ~52. The company is guiding for 65% growth in sales in the coming year, providing what I deem a reasonable value for the expected growth. The downward price momentum is strong, wouldnā€™t be surprised by a robust bounce back from being over-sold the last couple days, but I wouldnā€™t be surprised by further price erosion in the slightly longer term. They report earnings May 9th ā€¦ last earnings report was superb but didnā€™t stop the price drop, but it seems like combination of growth and earnings will put a floor under the price before too long, though it could be substantially below current levels.

One of the TA rules of thumb Iā€™ve heard in the past, which I donā€™t put much faith in but kind of watch anyway, is that stocks will tend to trade back through prior gaps. UPST today filled in some of a prior gap up from ~60 to ~80 back in March 2021. If the rule of thumb holds weā€™d expect UPST to trade back down to ~60. Which I donā€™t really trust, hence my buying today when it hit a level of valuation I could stomach. Iā€™ll buy more if it does go into the lower 60s.

5 Likes
2

I absolutely love, and I mean LOVE, the wonderful business of Upstart and its 8% loan origination fees. Surely the most sustainable business in world history!

2 Likes
3

Price action down has been heavily influenced by an ever growing portion of the float sold short. Over the last month, the short interest has grown from 17% to well over 20%. In spite of the garish statistic, that represents only about 1.5days of volume, so donā€™t look at too much of a short squeeze.

Another reference, MF darling and other hypergrowth issue, LMND (money losing insurance company) has 35% of float sold short.

I have been adding to my large position lately as well, although I started in the mid 80ā€™s.

3 Likes
4

I absolutely love, and I mean LOVE, the wonderful business of Upstart and its 8% loan origination fees. Surely the most sustainable business in world history!

If they can maintain an advantage more accurately qualifying loan applicants than their competitors and loan originators, then their business should be sustainable and grow rapidly. 8% loan origination fees certainly sounds steep, though, not sure how sustainable that precise number will be. These are for unsecured personal loans that carry a fairly high interest rate, and their fee will drop as they enter into collateralized auto loans and eventually mortgage loans.

They seem to have a big head start on building a high level AI/machine learning system for this purpose. It would take a large national bank undertaking a major multi-year effort to try to replicate it. Smaller regional banks and credit unions have no hope.

1 Like
5

Upstart is improving incrementally every day and their competition (FICO) is not.

2 Likes
6

possibly mess up their ratings, though such fears donā€™t seem to have much basis in reality as far as I can tell

Well, my understanding is the rating agencies accepting UPST ratings is critical, otherwise the banks will not be able to create MBS or ABS. Right now, rating agencies know about FICO ratings, but not about UPST rating methodology. I think the fear is warranted.

Just to keep an eye I have sold few $40 puts.

7

Upstart is improving incrementally every day and their competition (FICO) is not.

Why you say that?

8

Upstart AI models get trained with incoming data (~200 attributes).
As more data comes in daily, the models incrementally get smarter.
It is inevitable that few years from now the models are much better.

FICO scores are old and dumb but simple and entrenched.

1 Like
9

FICO scores are old and dumb but simple and entrenched.

If you are assuming that FICO is not doing AI/ML you are not properly informed. UPST is still gathering data and FICO is sitting with mountains of data. Data is as important, if not more than the model.

Separately, right now UPST is playing sub-prime credit. Trying to figure out better sub-prime risk is as old as wall-street.

To put it bluntly there are only so much gemsā€™ in tu$d. When they move to prime credit, there is not much value they can offer.

2 Likes
10

FICO scores are old and dumb but simple and entrenched

I think when you are looking at UPST, it may be beneficial to look at it independently and not as someone who is going to eat FICOā€™s lunch.

When in doubt after AMZN bought whole foods, I bought costco, but I also sold in the subsequent bounce. A costly lesson. Thinks donā€™t die as quickly as people think, just like turnaround takes times ( hehe IBM).

11

Separately, right now UPST is playing sub-prime credit. Trying to figure out better sub-prime risk is as old as wall-street.

To put it bluntly there are only so much gemsā€™ in tu$d. When they move to prime credit, there is not much value they can offer.

Yeah, that is a negative that isnā€™t well understood about UPST, there isnā€™t as much as most think there (both $ wise and population wise), and once you start moving up the FICO distribution there are better options available for most people. If I needed a car loan Iā€™d go to dealer financing or my CU, both of which you can get <3% and nominal fees if I needed cash Iā€™d use a HELOC or mortgage, again both of which are 3-6% and HELOC has essentially no fees and mortgage is tax deductible, great. To compete there you need to have large volume and a good relationship with a bank.

Maybe next they can get into payday loans.

Also I think AI is mostly dreams instead of performance, maybe at some point that will change. I do own some UPST though but donā€™t think Iā€™d buy here before earnings, these SAAS gets absolutely trashed if they donā€™t beat and raise every earnings. I donā€™t think Iā€™m selling out of laziness though, check back in a couple weeks and see how sorry I am.

1 Like
12

Also I think AI is mostly dreams instead of performance

Billions of people using Google or Amazon or Facebook are using AI/ML.

2 Likes
13

If you are assuming that FICO is not doing AI/ML you are not properly informed.

FICO is dumb because it uses very few data points. There is no need to force AI/ML when it is not needed.
It is a deterministic approach. It is entrenched and will take time to die slowly.

1 Like
14

"Separately, right now UPST is playing sub-prime credit. Trying to figure out better sub-prime risk is as old as wall-street.

To put it bluntly there are only so much gemsā€™ in tu$d. When they move to prime credit, there is not much value they can offer."

Yeah, that is a negative that isnā€™t well understood about UPST, there isnā€™t as much as most think there (both $ wise and population wise), and once you start moving up the FICO distribution there are better options available for most peopleā€¦

Upstart has already moved out of sub-prime for their unsecured personal loan and auto loan approval processes that they have available now. Of course thereā€™s not just ā€˜sub-primeā€™ and ā€˜primeā€™ but a whole spectrum of gradations. At the bottom end of the traditional credit rating spectrum, banks will typically not lend to anyone, but UPSTā€™s model/analysis finds a sizable minority of borrowers who are worth lending money to at the right rate. As you move up the scale towards prime, UPST finds more of those, and the traditional rating methods start to approve some of the same applicants, but UPST still approves a significant number of applicants that are not approved by the traditional process up to FICO scores of ~750: 4-5% of applicants with FICO 750 were rejected by traditional underwriting but approved by UPST in their study.

Furthermore, aside from the simple ā€˜yes/noā€™ of approval, thereā€™s the interest rate charged, which UPSTā€™s better risk assessment can help the banks with, and individual customers might see offers with higher or lower interest rates than from a bank using traditional underwriting. This will apply to a broad swath of applicants in the good but not great FICO categories.

Graphs showing the curves of those numbers, and a lot of other good information, are available in this study: https://www.nber.org/papers/w29840

"
Invisible Primes: Fintech Lending with Alternative Data

[Abstract]
We exploit anonymized administrative data provided by a major fintech platform to investigate whether using alternative data to assess borrowersā€™ creditworthiness results in broader credit access. Comparing actual outcomes of the fintech platformā€™s model to counterfactual outcomes based on a ā€œtraditional modelā€ used for regulatory reporting purposes, we find that the latter would result in a 70% higher probability of being rejected and higher interest rates for those approved. The borrowers most positively affected are the ā€œinvisible primesā€ā€“borrowers with low credit scores and short credit histories, but also a low propensity to default. We show that funding loans to these borrowers leads to better economic outcomes for the borrowers and higher returns for the fintech platform.
"

If I needed a car loan Iā€™d go to dealer financing or my CU, both of which you can get <3% and nominal fees if I needed cash Iā€™d use a HELOC or mortgage, again both of which are 3-6% and HELOC has essentially no fees and mortgage is tax deductible, great.

Yeah, if you are at the very top of the scale on the traditional model, then UPST probably isnā€™t going to help you, since youā€™re already getting all the best offers. I suppose there are some exceptions to that rule where UPSTā€™s AI gives a better risk assessment on even the top rated FICO scores that would benefit either the lender or the borrower somehow, but thatā€™s not ever going to be where their bread is buttered. Of course thereā€™s only about 30% of the population with FICO scores over 775, where thereā€™s virtually no disagreement between FICO and UPST, so that leaves a whole lotta population out there where some percentage will benefit from UPSTā€™s analysis.

Maybe next they can get into payday loans.

Next up in their roadmap going forward are small business loans and small dollar loans. I guess the small dollar loans are the same market as payday loans?

2 Likes
15

I do own some UPST though but donā€™t think Iā€™d buy here before earnings, these SAAS gets absolutely trashed if they donā€™t beat and raise every earnings. I donā€™t think Iā€™m selling out of laziness though, check back in a couple weeks and see how sorry I am.

I bought the other day because it hit a valuation level I could stomach. Predicting when the stock will find a bottom is above my pay grade. Itā€™s highly volatile and if they come out with another killer earnings report, maybe it takes off again and doesnā€™t look back. Or maybe it keeps going down even with a great earnings report like last time. Or they have a crummy earnings report and it really craters.

Iā€™ve spent a while reading up about the company and thinking about it, and I definitely buy into their growth story/opportunity, management team, and a huge addressable market that is critical to the economy. I also love that their product provides a big win/win for both the consumers they are qualifying for a loan who couldnā€™t get it otherwise and the lenders who find incremental profitable business they didnā€™t have before.

I think the stock has dropped to a level that your comment ā€˜these SAAS get absolutely trashed if they donā€™t beat and raise every earningsā€™ doesnā€™t really apply. UPST isnā€™t valued like a SAAS stock anymore. They are profitable and trading at ~52 PE and 7.5 times sales using trailing twelve months data. Most SAAS stocks that Saulā€™s board follow donā€™t have GAAP earnings and so you canā€™t calculate PE, and the price/sales is double-triple UPSTā€™s or more.

UPST is in that awkward phase transitioning from being valued as a go-go growth stock (to the moon!!!) to being valued on its actual earnings. They have guided for ~65% growth in sales this year, Iā€™m okay paying 52 PE for that, given the size of the opportunity open to them.

Doesnā€™t mean the stock canā€™t drop another 50%, though! As I said in my OP to this thread, thereā€™s some TA reason to expect it to drop to ~$60, another 20%, and I wouldnā€™t be surprised by that, nor would I bet on it. If it does drop that far without terrible news breaking, Iā€™ll buy more. If it drops in half to ~$37, without terrible news breaking, Iā€™ll buy more again, though I do doubt it would go that low. PE 26 for this kind of profitable growth stock seems way too low, but it could happen in a market panic.

16

I do own some UPST though but donā€™t think Iā€™d buy here before earnings, these SAAS gets absolutely trashed if they donā€™t beat and raise every earnings. I donā€™t think Iā€™m selling out of laziness though, check back in a couple weeks and see how sorry I am.

This is what I get for not taking my own advice, 2% growth QoQ for a supposed hypergrowth stock, lowered guidance to about what it was last year.

Mostly I was just being lazy and not selling because I didnā€™t have a good idea where to put the money and it was already beaten down so there was some hopeā€¦ I feel like I need to get over my fear of cash (hopefully I can do this before UPSTā€™s next earnings call).

2 Likes
17

In my many decades of life Iā€™ve never seen such euphoria as Saulā€™s board for UPST. UPST is the least sustainable business in world history.

18

Mark Minervini went to CNBC to pump his UPST, around Oct 2021, while not able to recollect what UPST did, he acted as if he didnā€™t hear the question and zoom glitch, but he could not recollect the name. That market the top of UPST. When people are going in front of camera to declare I own the stock without even knowing what the heck the company doesā€¦

We have some ways to goā€¦ Everyone their mother, and mother-in-law should be screaming UPST has magic formula to make sub-prime loansā€¦

19

UPST is the least sustainable business in world history.

Hi chompin,

Iā€™m a fan of your posts. I was wondering if you could flesh out why you are so negative on UPST? They are basically loan brokers, and certainly mortgage brokerage has been a fairly large, sustainable business for a long time. Maybe not so much the lower value personal loans and auto loans that UPST has been working on so far, but I donā€™t see why the concept wonā€™t sustainably work if they can gather applications in mass volume and have an improved loan qualification process, both easier for the applicant and more effective at quantifying risk.

I guess you donā€™t think they have any kind of sustainable edge in gathering loan applications/approval?

I bought some more today at $29.8-30.2. Trailing twelve month PE of 19 at that price. Of course the stock crashed because they lowered their guidance to basically flat revenues through this year, so all the people who were counting on more growth this year bailed. I still like their business model and growth prospects for the future along with the earning power theyā€™ve demonstrated, but the guidance to virtually no growth this year is obviously a concern.

4 Likes
20

I bought some more today at $29.8-30.2.

If you are planning these buys with long-time horizon, then there is no need to rush wait for the dust to settle. If you think sell off is excessive, then be prepared to immediately book your gains.

In either case, after such a steep sale, expect further volatility.

It is difficult but I would ask myself, since you liked the company at much higher prices, is your thesis wrong, or market is wrong?

Sometimes, accepting losses is the best course of action. I know my ego doesnā€™t allow me to accept losses when they are small, and after it is truly crushed with a severe loss, I accept it.

2 Likes