You Always Remember Your First: Shit Quantification
You Always Remember Your First: Shit Quantification
Closing the loop on a shitty Chinese tech company.
This is the final part of a two-part series called “You Always Remember Your First”, a tale of teasing apart a shady, publicly traded, Chinese technology company. Read the first post here if you haven’t already.
With all of our QTT background research, our initial technical analysis, perpetual bad attitude and a Coke Zero, it’s time to try to tie the application data to QTT’s revenues.
In QTT’s filings, we can see that the QTT application makes its money from selling advertising to other application developers. Therefore, we decided to focus on the advertising ecosystem for other applications.
So we need to monitor (or what we call ‘instrument’) the QTT application running on a phone, observe the network traffic flowing out and back from their servers and then start to make sense of what we are seeing.
On Technical Analysis and Sample Sizes
When you are analyzing a high-tech company, be it a news reader, an online gambling platform, or for example, a shitty online car retailer, you need to be able to collect a meaningful enough sample without doing anything evil or being overly aggressive.
While I possess no math background, 1,000 samples is generally considered a statistically relevant sample size. On the upper threshold, I’ve often been advised to aim for 10% of the overall population size but try to stay within 1,000.
We usually use this as a baseline, but we often aim to collect significantly more data to ensure that we can test and backtest our hypotheses.
However, the trick is to collect this data passively as much as possible.
For example, there are times when you have to web scrape or ‘poll’ a web system’s API to retrieve information you wish to process. Being aware of how you do this polling and how much polling you do is critical, as being over-aggressive can do harm and not gathering enough data can be statistically misleading. This is considered an active collection in our world.
Real creativity generally happens when you, as a technical analyst, can devise ways to passively collect as much information as possible without generating any traffic to the target system outside of the applications normal usage. For example, by scrolling around the QTT application while observing its network traffic, you can begin to capture advertisements, analytics trackers, and other information that does not require us to do anything active besides using the application.
This is as much a legal discussion as a technical one, but remember, the more data you collect, the more reason you need to explain why you did so.
Now, generally speaking, once you have retrieved a sample, you need to use a spreadsheet (ideally) or a data visualization tool to interpret the data.
This brings us to the next critical point that I would like to highlight for you, dear reader.
Bullsh*t Hunting is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
Visualization Doesn’t Have to Be Fancy to Be Good
You will hear me wax poetic endlessly about my desire to make data accessible and to polish the user experience of exploring data. People still don’t listen to me on this, at their own peril.
A prime example is how we solved this first phase of QTT with the world’s dumbest visualization.
No, not some fancy graph, not a big map, no algorithms, no fancy shit.
Think cave-dweller visualizations.
I built a big-ass HTML page that outputs the ads, their landing page and the advertiser’s application identifier. So that we can scroll steadily and let our eyes and brains do the pattern matching on their own.
If your users have to spend 15 minutes screwing around loading data into a platform, cleaning it or otherwise watching a YouTube tutorial on how to actually use your tool? You’ve just burned my most valuable analytical energy.
The initial analytical energy that comes at a massive investigative cost.
So what did we see when we pumped this all out to HTML? A bunch of poorly-Photoshopped, repeating advertisements for…well, junk.
What immediately interested us was that advertisements such as the ones above were all associated with the same apps: toothbrushes, movies, massages, games, and… dating?
Not generally things you would all see pointing to the same applications, not even Amazon has dating (I think?).
Many of the advertisements destined for the same cluster of Android application identifiers (such as "com.mengtuiapp.mall"). I kept scrolling and scrolling.
Over and over, I kept seeing the same patterns.
Something was really strange here that either meant we had a data collection problem, where I needed to get a wider sample of advertisements, or we were seeing a very unhealthy advertising environment.
Or worse.
I took the top advertisers' Android application IDs, tracked down each company on the Google Play store, and sent the spreadsheet to a fluent Chinese-speaking analyst, letting them know I'd be interested in finding out who these companies are.
Then I went back to work, trying to tease apart just how their application works and who the hell would actually want to use it.
And I waited.
The Big Moment
A few days later, I was brushing my teeth on the top floor of our rented house in the south end of Saskatoon, Saskatchewan, when the phone rang. I saw that it was one of the analysts calling, so with a mouthful of Crest Complete, I answered it on speakerphone and mumbled around the mint: "Wuffs up! Find anything?"
Before I can clean the toothpaste spit bubbles off of my iPhone, the analyst starts excitedly yelling at me that we cracked the case.
Those top advertisers that we had discovered? By running further Chinese-language due diligence, the analyst discovered that these companies were privately owned by none other than: Eric Tan, CEO of QTT himself.
She was spittin’ bars. I was spittin’ toothpaste.
It was a whole-ass scene.
We were ecstatic.
What the analyst was telling me, and what I'm telling you, is that QTT was using their platform as a way to suck money in one end (from investors, potentially you), enrich themselves from that investment money, and then further enrich themselves personally by pumping ads out for apps that QTT's CEO, his wife and homies owned.
These other apps now give Eric Tan further opportunities to monetize downloads, advertising and other methods of churning revenue, all the while back-feeding and buffering the main QTT app as having a "healthy advertising base." It doesn't matter if QTT fails or goes bankrupt because he and his inner circle are making the same amount of money whether it is wildly profitable or not.
Perfect scheme.
In fact, in our samples, over 25% of the advertisements pointed to Meng Push, an application ultimately owned by Tan and his wife.
See how it works? Big old circular clusterfuck. And remember, when you're talking about hundreds of millions of dollars in alleged (we say alleged revenue for a reason, keep reading) revenue, that's a lot of clusterfuckery.
It would be like if the only advertisements on Twitter you ever saw were for Tesla and SpaceX, right? The horror. It's bad enough as it is.
We knew we were on to something now, but we were still far from having good, solid evidence that something bad was happening in China. We had circumstantial due diligence, and now it was time to close the loop fully.
Closing the Loop
Now, like in any intelligence or investigative operation, you have to close the loop on these findings. We had very good circumstantial evidence that something bad was going on: there were undisclosed related parties, the company was improperly pumping advertisements, and Mr. Tan was making enough money off of this to buy that sweet, sweet pad in Cali.
However, if you are a good activist short seller, more work is needed to prove it to the point where the market might pay attention to you and meet strict regulatory and legal standards.
So you usually want to go right to the ground.
In this case, Shanghai, where investigators, at great risk, went out and took photographs of these alleged advertisers and application developers. What they found was pretty shocking.
Four companies registered to Tan were in an empty warehouse:
And run-down, empty office buildings serving as nothing more than virtual mailing addresses for Tan’s network of companies:
The on-the-ground due diligence included pulling local Chinese financial filings and discovering, at the time of the report in 2019, over 69 active lawsuits against QTT or its principals, primarily for intellectual property theft—a whole other angle we didn't even get into today.
It was incredibly valuable, it told us that our technical analysis was the compass that pointed to the real evidence of fraudulent activity going on in China.
Further Chinese documents revealed they were faking how much cash they actually possessed when they were filing documents with the SEC.
This had all the hallmarks for an activist short seller to step up and call bullshit.
Wolfpack Research published a report on September 12, 2019, with reams of data and analysis on the ground investigation, supported, in part, by my work examining numerous technical and data angles.
I was excited to see what happened.
So? What happened?
Nothing.
No one gave a shit. The stock bounced around a bit. There was a tiny bit of news and a bit of back and forth between Wolfpack2 and QTT3. Then, it all just fizzled out.
I was pretty dejected but not dissuaded. I was convinced. We had done the work and presented the data and our case, and sometimes, you still just don't win.
But were we wrong, not from an investment standpoint (yes, because they continued to trade) but from an investigative standpoint?
You make the call.
$QTT continued to plummet in value until July 2020, when a Chinese news outlet, CCTV4, ran an expose that showed QTT was running a potentially illegal advertising platform. The Chinese government decided to investigate.
After the news, and throughout 2020, the stock plummeted, leaving the insiders rich and the shareholders fucked.
At the time of this writing, QTT has been delisted and no longer trades in North American markets.
They've faced a handful of lawsuits since 2019.
“QTT’s Top 12 Lies and Omissions for the New Year”. Wolfpack Research. January 17, 2020. https://wolfpackresearch.com/qtts-top-12-lies-and-omissions-for-the-new-year/
“QTT Responds to False and Misleading Report by Wolfpack Research”. PR Newswire. December 27, 2019. https://www.globenewswire.com/news-release/2019/12/27/1964629/0/en/Qutoutiao-Inc-Responds-to-the-False-and-Misleading-Report-by-Wolfpack-Research.html
“Qutoutiao Tanks After CCTV’s Consumer Rights Show Targets News Reader App” - https://www.yicaiglobal.com/news/cctv-gala-fells-qutoutiao-shares-23-prompts-app-stores-to-unshelve-it