Does UPST make the cutoff to rejoin my portfolio today, after reporting great Q4 numbers?
As a recap: I first bought UPST in May 2021 after Q1 earnings and maintained at least 80% allocation (90% on earnings day through Q2 and Q3). Why did I go crazy with the allocation? At the time I was overly confident with looking at review counts, google trends, website traffic data to correlate with earnings results, after ‘backtesting’ it to prior quarters both pre and post IPO. It worked fantastically for Q2.
Obviously, fraud attack in Q3 threw all that out the window. I sold to 0% allocation on Q3 numbers, retook a 15% allocation the next day, then sold all of that again, in first week of December 2021 when I felt the valuations for superior hypergrowth SaaS companies were too attractive to pass up.
No real positions were re-established since then, particularly after my other concerns of UPST’s business mounted.
Now, post Q4 – how did UPST do?
First, Q4 results.
(I wrote on twitter my prediction/expectation yesterday)
Expectation: 494K loans, $285M fee revenue, avg loan size $8600 (flat QoQ), fee take rate 6.7%
Actual: 495K loans, $287M fee revenue, avg loan size $8275 (down QoQ), fee take rate 7.01%
Q4 met my prediction and expectations perfectly for fee revenue and loan count. Loan sizes fell more than I like, but was made up for by fee take rate increase.
Why did I want loan sizes to stay flat, even though we know they are moving downmarket to lower, ‘traditionally subprime’ borrowers who tend to take smaller loans?
Because they talked about being competitive for traditionally higher FICO, higher income, larger loan borrowers last quarter. I want to see these ‘bigger borrowers’ make up for the smaller fish. Looks like that might not necessarily be happening (yet?).
So, loan size drop remains a problem, but in this context it’s a minor issue, as take rate did go up to compensate.
I’m not sure what’s driving the fee take rate change. The 10Q might provide information on this, and hopefully this is something that is sustained for next quarter.
Second, FY22 guidance.
Expectation: Q1 guide $320M, FY22 guide $1.1B with no auto guide given.
Actual: Q1 guide $305M, FY22 guide $1.4B with “$1.5B auto financing volume”
This is a huge upside surprise. $1.4B total rev guide is way bigger than I expected. If we assume they ‘beat and raise’ as necessary for the wall-street game, actual FY revenue may be higher.
What do I want UPST to do in FY22 (and I have to believe that they can realistically do it)?
This answer for me is dependent on what a non-SaaS company needs to grow, to be considered “on par” with a SaaS-like company.
Non-SaaS, direct consumer facing business will command lower valuation than a “recurring revenue-like” B2B company.
As an example, any future economic slowdown will curtail UPST more than BILL.com (see UPST’s -80% drop QoQ in COVID vs BILL’s +5.5% QoQ growth, as proof).
Lumpy ‘unexpected’ “growth volatility” quarter to quarter will also be lower valued (this is certainly less desirable than a company with merely an ‘expected’ quarter to quarter seasonality).
So, in order for me to consider UPST on par with other companies I already own, I want to see FY22 deliver >20% QoQ growth, every quarter.
As I wrote in prior posts/tweets, in 2021 pre-Q3, I was already expecting at least 30% QoQ of fee revenue growth, so this is nothing new – it’s not a goal post being moved here.
Last year, I expected UPST to taper down to >20% QoQ through 2022, as the law of large numbers kicked in for personal loan growth.
That means I want something like: Q1 345M, Q2 414M, Q3 497M, Q4 596M which equals 1.85B in fee revenue. Add in interest-income (the low quality revenue that I don’t care about), that pushes it to $1.96B in total revenue.
Does that number seem “insane” to you? Well, in 2021, it didn’t look too crazy to me, prior to Q3’s failure! It’s why I didn’t sell at $400 at the “top” months ago – at the time, I fully expected this to be realistic to achieve.
Is it possible today? I’m leaning towards ‘no’.
Assume loan sizes continue to fall 4% each quarter (which matches the current trend) and take rate stays at 7%. This means to generate 1.85B in personal loan fee revenue, 3.6 million loans must be originated. It means next Q1, they need to do 620K loans to be on track for this.
Trustpilot review count appears to correlate quite well with loan transaction numbers. (It’s how I arrived at Q4’s ‘near-perfect’ loan transaction number prediction)
6080 reviews in Q3 with 362780 loans. That’s 59.7 loans per review.
8336 reviews in Q4 with 495205 loans. That’s 59.4 loans per review
At roughly halfway through Q1 (February 15 2022), there were 4236 reviews.
4236 x 2 x 59.55 = maybe 504507 loans for Q1. Unfortunately, that falls short of the 620K needed.
Well, this now brings me to the next item.
Third, Q1 guide.
504507 loans at identical take rate/loan size = 292M fee revenue. Add interest income, maybe 309.5M total revenue in Q1.
That’s an in-line result with a tiny beat above their Q1 guide of $305M. (This matches well with their typical Q1 seasonality)
This is why auto revenue is absolutely critical. Since I can’t realistically expect UPST to do 1.96B total revenue in 2022 off of personal loans alone with the above calculations, we need auto fees to kick in serious high gear to counteract the law of large numbers.
Fourth item: auto guidance.
They guide for FY22: $1.5B in auto loan financing volume
This is a disturbingly small guide. If we assume a “take rate” of 3%, that’s auto revenue of $45 million. Not enough to make up for a personal loan shortfall, even if their ‘actual’ result ends up double that of this initial (presumably conservative) guidance.
For the first time, they also did not give out auto-refi transaction numbers for Q4, which is a ‘red flag’ to me. I have a suspicion they would be bragging about a huge jump in auto-refi transactions, had it actually occurred.
Finally, it remains very vague what exactly auto revenue economics are like. Per the conference, they will be getting revenue ratably over the life of loan or something? and no upfront fees?
Other concern: delinquency rates
I find it funny they cited our “public forum” regarding delinquencies, when in fact, the person who first noted the rising delinquencies as a problem was the Wedbush analyst, not me, in December (who then flagged a continued delinq spike in January a second time). My post about it was in January.
Regarding their explanation of delinquencies: I am no finance expert so I can’t do anything but take their claim at face value that everything is perfectly fine. Hopefully things are.
The best proof of their performance, of course, is with time, so it is still important to monitor the KBRA reports.
The Feb 4 surveillance report (which I don’t think anyone posted about before here), did NOT look great for 2021 trusts (for the first time, UPST pass throughs’ cumulative net loss rates were under-performing the KBRA projections! See column 2 in the table vs column 3: https://imgchest.com/p/ne7bppjk453)
A minor ‘quibble’ about share buyback
This is somewhat of a turn off. Why announce a $400M share buyback authorization to signal confidence in your shares being underpriced, when instead, management could have just stopped selling their own shares at the lows, and personally buy them on the open market?
For example, the CEO of Asana has been doing this (see here: http://openinsider.com/insider/Moskovitz-Dustin-A./1549917)
Meanwhile, here is UPST’s CEO sales since Q3: http://openinsider.com/insider/Girouard-Dave/1832805)
I mean, come on, if I truly believed my company is underpriced, and believe my business is destined to become a FAANG, would I really be trimming my stake like this?
At the very least, he could have been like Paul Gu, who did stop selling entirely, after the share price crashed post-Q3 (http://openinsider.com/insider/Gu-Paul/1832812).
If I had a portfolio of 20+ companies, I’d buy UPST here. But I’m running a concentrated hypergrowth basket, so if I want to include something it has to meet really high expectations. I need to be convinced to sell parts of my other high conviction holdings for UPST.
For a non-SaaS company with outsized macro risks, I’m looking for all metrics to be ‘perfect’. The way I frame it for myself: Why should I buy a company that has lumpy, not recurring revenue, if they are not significantly exceeding the growth rates of a SaaS-like company?
I would choose DDOG at 70% growth in 2022 over UPST at 90% growth in 2022, for example.
DDOG and my other portfolio companies also do not have similar regulatory risk, concentration risk, recession risk, fraud risk etc as UPST.
-Q4 result met my expectations
-FY guide greatly exceeded my expectations. Guidance 1.4B (And I want the actual results to be over 1.9B)
-FY auto guide of 1.5B transaction volume however, is way too low to me. A better auto guide with more clarity, would have pushed me to add UPST. I can’t see them getting to 1.9B total revenue easily, without better auto loan visibility. It’s interesting they don’t mention autorefi transaction volume on yesterday’s call, for the first time.
-Unclear picture on autorefi or auto-retail take rate - they will be getting revenue ratably over life of loan or something? and no upfront fees?
-Delinquency rates still remain a concern to me - there are vintages for the first time that are underperforming KBRA projections in February
I also want to say this loud and clear: this is the very high bar I have set for my own portfolio. Your portfolio thresholds are likely very different.
Best of luck to all shareholders and I hope UPST blows it out the water for quarters to come!