Latest / Deep Values: Value Investing Deep Dives / PDD‘s Q2 Earnings Financial Analysis
Transcript
- 0:00If you want to know the future of a multi billion dollar
- 0:02company, you don't look for a crystal ball or or you know,
- 0:07wait for some highly paid pundit on a financial news network.
- 0:10You actually look at, well, line 42 of a Python script.
- 0:13Yeah, that's where the real truth is.
- 0:15Which is crazy because we tend to expect financial prophecy to
- 0:19look, I don't know, polished. We exect these 100 age white
- 0:25papers filled with glossy charts or.
- 0:27Flick resentations. Exactly the ones designed to
- 0:29sell you a narrative. But today our source material
- 0:33strips all of that away. We are looking at just raw
- 0:36unadulterated Python code. Raw code.
- 0:39Yes, specifically a script titled the PDD Financial
- 0:42Evaluation Script and DCF Analysis Report.
- 0:45OK. And we are staring directly at
- 0:47the underlying math, trying to translate, you know, variables
- 0:51and loops into a coherent narrative about a single
- 0:54company's financial destiny. Because reading a script like
- 0:57this is essentially reading the bare metal DNA of an analyst's
- 1:00thought process. You're seeing the exact
- 1:03mathematical weights they assign to risk to time to value.
- 1:07And it's completely stripped of any marketing spin or or
- 1:10corporate PR. Exactly.
- 1:12It is just logically bare. But before we start dissecting
- 1:15that logic, it is really vital to establish what we are
- 1:19actually looking at, yes. So disclaimer time, the data and
- 1:23analysis embedded in this script, it is based entirely on
- 1:26publicly available sources, right?
- 1:29And it may be subject to revisions or differing
- 1:31interpretations. All contents from this channel
- 1:36is for research and entertainment purposes.
- 1:38Only right? It does not constitute any
- 1:40investment or financial advice or service.
- 1:43You have to conduct your own due diligence before making
- 1:46investment or financial decisions.
- 1:48And for those of you who do want to take that research to the
- 1:51next level after we finish here to access more advanced
- 1:53researches, follow Guru's latest movement and leverage our latest
- 1:56Deep Values agents. Welcome to download Deep Values
- 1:59APP from App Store. Highly recommend that.
- 2:01Yeah, and to track stock market temperature.
- 2:03Welcome to download iOS app Buffett indicator from Apple App
- 2:06Store or Bubble Watch from Google Play and links are in the
- 2:10description. Because having those tools at
- 2:12your disposal is incredibly useful.
- 2:14Yeah, especially when you start understanding the mechanics of
- 2:17how these financial models are actually built in the real
- 2:19world. Right?
- 2:20And today we are tearing down the architecture of one of those
- 2:23models. We're looking at the engine, not
- 2:26making stock pick. Exactly.
- 2:27And Speaking of the engine, the engine of this entire script is
- 2:31this one primary Python function.
- 2:34It's named def run under score DCF.
- 2:37DCF. Right, which stands for
- 2:39discounted cash flow. And for anyone who isn't, you
- 2:43know, spending their days building financial models, we
- 2:45should probably clarify what that actually means.
- 2:47Yeah, we should. So the code takes 5 very
- 2:50specific inputs and the first one is SCF under score 0.
- 2:56The starting free cash flow. Right.
- 2:58And let's pause right there, because when we talk about free
- 3:01cash flow, we aren't just talking about revenue.
- 3:03No. Not at all.
- 3:04Like it's not just the money coming in the door.
- 3:06Right, because a lot of people confuse that.
- 3:08They see huge revenue and think massive success.
- 3:11But we are talking about the actual usable cash a company has
- 3:14left in the vault after paying for all of its basic operating
- 3:17expenses and necessary capital expenditures.
- 3:19The vault. It's the ultimate measure of
- 3:21financial health. Yeah.
- 3:23I mean, a company can have massive revenue, but if its
- 3:26expenses are equally massive, then what?
- 3:28It's free cash flow might be 0. Wow.
- 3:31Yeah, the free cash flow is the surplus.
- 3:34It is the money the company can actually use to pay dividends to
- 3:38investors, or buy back stock or aggressively expand into new
- 3:42markets. So it's the real money.
- 3:43Yes, that is the baseline number this entire script revolves
- 3:47right around. OK, let's unpack this then.
- 3:49The script provides A baseline variable for this.
- 3:52It says SCF under score TTM. Which is trailing 12 months,
- 3:56right? Trailing 12 months of free cash
- 3:57flow, it sets it at 14,850. So we have our starting pile of
- 4:01cash, OK. And the function then takes four
- 4:03other arguments. It takes growth rate, discount
- 4:06rate, terminal growth rate and shares out right.
- 4:09And the script executes a very specific command, it just reads
- 4:12for under score in range 10. A10 year loop.
- 4:15Exactly. It runs a loop exactly 10 times,
- 4:18Yeah. And in each loop it takes that
- 4:21current free cash flow and multiplies it by 1 plus the
- 4:24growth rate. So it is mathematically
- 4:26projecting the cash into the future.
- 4:29What's fascinating here is that specific loop is the mechanical
- 4:33lever of compounding interest. OK, it is projecting how a
- 4:36single cash pile today behaves over a decade under specific
- 4:41conditions. You know, my brain instantly
- 4:43visualizes this as a as a highly constrained financial time
- 4:47machine. I like that.
- 4:48Yeah, like we take today's cash pile of 14,850 and we projected
- 4:52into the future year by year compounding it, but we only
- 4:56travel exactly 10 years into the future.
- 4:58Only 10 years. Right, and why 10?
- 5:00Like if an analyst is building a time machine in Python, why not
- 5:03project out 30 or 50 years to see the total lifespan of the
- 5:07company? Well, the 10 year constraint
- 5:08represents a fundamental acknowledgement of the limits of
- 5:11human predictability. Oh, OK, because forecasting a
- 5:14company's performance one or two years out is incredibly
- 5:17difficult. Sure, predicting five years out
- 5:19is bordering on speculative fiction.
- 5:22Speculative fiction. I love that.
- 5:23It really is, yeah. And by the time you reach year
- 5:2610, you are stretching the absolute limits of mathematical
- 5:30logic. Because things just change too
- 5:32much. Exactly.
- 5:34Think about the variables in the real world.
- 5:36Over a decade, entirely new technologies disrupt whole
- 5:39industries. Consumer habits fundamentally
- 5:42shift. Global supply chains
- 5:45restructure. The analyst caps the loop at 10
- 5:48because beyond a decade, projecting specific year over
- 5:52year, operational growth becomes entirely meaningless.
- 5:55Wow, Yeah, I guess you you can't model for the invention of the
- 5:59next smartphone or some sudden global shift.
- 6:01And the basic financial theory here is that a dollar today is
- 6:06worth more than a dollar tomorrow because of, well,
- 6:10inflation and opportunity cost. Like if I have a dollar today, I
- 6:13can invest it and earn a return, but if you promise me $1.10
- 6:17years from now I lose an entire decade of investment
- 6:20opportunity. The opportunity cost is a huge
- 6:22factor there, but that 10% discount rate also quantifies
- 6:26risk. Every single year you project
- 6:29into the future, the risk of the company will fail to generate
- 6:32that cash increases exponentially.
- 6:35Oh, I see. So the discount rate is the
- 6:37hurdle the company has to clear, right?
- 6:38The analyst is essentially saying if you want me to assign
- 6:41value to this theoretical cash in year 8 or year nine, you have
- 6:45to compensate me for the immense risk that your company might
- 6:48stumble. Or face new competition.
- 6:50Exactly or completely fail to deliver.
- 6:52It's systematically shrinks those massive compounded future
- 6:55numbers down to a much more sober present day reality.
- 6:59So if that 10 year loop is the baseline engine, the next
- 7:04logical step is seeing how an analyst tries to break their own
- 7:07engine. Because the script doesn't just
- 7:09run one timeline, it maps out multiple scenarios.
- 7:12Multiple universes basically. Yeah, and the first two
- 7:15scenarios are entirely focused on stress testing the company
- 7:18with stagnation and decline, the analyst forces the model to
- 7:22digest deeply pessimistic assumptions.
- 7:24Well, stress testing is where you separate robust companies
- 7:27from fragile ones. You have to model the dark
- 7:29timelines. Like the dark timelines, right?
- 7:30And here's where it gets really interesting.
- 7:32So the first stress test is is labeled no growth, OK, And it
- 7:35calls the run DCF function and inputs a 0.0 for the growth
- 7:40rate. Makes sense, right?
- 7:41And critically, a 0.0 for the terminal growth rate.
- 7:44OK, we'll get to that terminal rate in a moment.
- 7:46But the second scenario is where the math gets genuinely
- 7:50intriguing. It is labeled conservative.
- 7:52Let's hear it. So instead of plugging in a
- 7:54round generic number for a downturn like, you know -5% or
- 7:58-10%. Which is standard.
- 8:00Right, very standard. The script passes in a highly
- 8:04precise -.0909 growth rate, and the code even contains an
- 8:10embedded comment next to it. It says -9.09% drop from 2025
- 8:16yo. Why?
- 8:17When you see a four decimal point level precision like
- 8:20-.0909 in a valuation script, it signals a complete departure
- 8:25from generic forecasting. Yeah, that level of precision
- 8:28makes me think this coder isn't just, you know, throwing darts
- 8:32at a board. Not all.
- 8:33They are translating a very specific known external threat
- 8:38into a mathematical input. Yes, because a generic -10% is a
- 8:42guess, right? But -9.09% implies they're
- 8:45calculating the exact known cost of maybe a specific tariff
- 8:49taking effect in 2025. Or they're modelling for the
- 8:52precise revenue loss from a major contract that's expiring
- 8:55next year. Exactly.
- 8:57It's like they're looking at a specific storm on the horizon
- 8:59and they're calculating the exact wind speed.
- 9:01They are mapping an external real world anxiety directly into
- 9:05the core logic of the model. Wow.
- 9:06That precise number is likely the output of an entirely
- 9:09separate internal model. Yeah, that calculated the exact
- 9:12impact of a macroeconomic contraction or a specific supply
- 9:16chart bottleneck they anticipate in 2025.
- 9:19South, this script is just the the final destination for a lot
- 9:22of other math. Well, even more revealing than
- 9:25the 9 point and 9% drop is the other 0 coded into both of these
- 9:31pessimistic scenarios. The terminal growth.
- 9:33Rate, terminal growth rate in both the no growth and the
- 9:35conservative scenarios, the terminal growth rate is absolute
- 9:3800.0. Now, I understand year to year
- 9:42growth being flat or negative in the short term.
- 9:46You know during that 10 year loop we talked about, sure.
- 9:48But terminal growth is the assumption of what the company
- 9:50does forever in perpetuity after the 10 year loop ends.
- 9:55This raises an important question, because setting the
- 9:58terminal growth rate to absolute zero forces A reckoning with one
- 10:02of the most deeply held assumptions in modern
- 10:03capitalism, which is the assumption of perpetual
- 10:06expansion. Because in almost every
- 10:08mainstream financial model, companies are just expected to
- 10:11grow indefinitely. Line goes up.
- 10:12Forever, right? But a 0.0 terminal growth rate
- 10:16forces the model into a reality where the company simply stops
- 10:21expanding its footprint. Yes, but what does that actually
- 10:24look like mathematically? Well, it means the company
- 10:26transitions into a steady state utility.
- 10:29But utility, yeah. It doesn't go bankrupt, but it
- 10:32never captures a new market again.
- 10:34It churns out the exact same nominal amount of value year
- 10:38after year, forever. OK.
- 10:40And when you factor in macroeconomic inflation, a 0%
- 10:44terminal growth rate actually means the company is slowly
- 10:47shrinking in real purchasing power over the decades?
- 10:50That is so bleak. It is.
- 10:52The market is so conditioned to tech companies promising
- 10:55exponential infinite growth, right?
- 10:58So a 0 terminal rate requires the investor to visualize a
- 11:02reality where a company peaks, fully saturates its total
- 11:06addressable market, and then just exists in carbonite.
- 11:09It's frozen in carbonite. It's it's like the equivalent of
- 11:11a massive blockbuster movie franchise that completely stops
- 11:15producing new sequels and spin offs.
- 11:17Great analogy. And they just rely entirely on
- 11:19the streaming revenue of the original trilogy forever.
- 11:22Yes, it continues to generate cash, but its cultural and
- 11:25financial footprint is just completely frozen.
- 11:27And mathematically, applying a 0 terminal growth rate acts as a
- 11:31massive anchor on the final valuation of the company.
- 11:35Really. Yeah.
- 11:36In a discounted cash flow model, the terminal value that forever
- 11:39period starting in year 11 often accounts for well over half of
- 11:43the company's total calculated present day worth over half.
- 11:46Because the future is where the bulk of the theoretical value
- 11:49lives, right? If you drop the growth of that
- 11:52forever period to 0, the final valuation craters.
- 11:56It is a ruthless, punishing stress test of the company's
- 11:59current price. Wow.
- 12:01OK, so 0% means a slow frozen decline against inflation.
- 12:05We really need to look at what the math of surviving and
- 12:08thriving actually looks like in this script.
- 12:10The sunny timeline. The sunny timeline, yes, the
- 12:12script provides 1/3 scenario to contrast the pessimism and it's
- 12:15called nominal. OK.
- 12:17And the inputs here paid a drastically different picture.
- 12:19So the nominal model calls the run DCF function and projects a
- 12:2415% growth rate, so .15. That is a lot.
- 12:27That is 15% growth compounding year over year for ten straight
- 12:31years. And crucially, the terminal
- 12:35growth rate is restored. It actually pulls the baseline
- 12:39variable from the top of the script, which is point O2 5 or
- 12:422.5%. Sustaining a 15% compounding
- 12:45growth rate for an entire decade is an incredibly rare feat in
- 12:49corporate history. Yeah.
- 12:51I mean, just to put the math in perspective, yeah, if you take a
- 12:54cash pile and grow it by 15% every year, compounding for 10
- 12:57years, you effectively quadruple the size of the company.
- 13:01Exactly. It requires relentless
- 13:02execution. Aggressive market capture?
- 13:04Yes. Continuous product innovation
- 13:07and you have to fend off every competitor for a decade
- 13:09straight. Which almost never happens.
- 13:11Right. But then after that grueling 10
- 13:14year Sprint, the script caps the forever growth rate at 2.5%.
- 13:18Yeah, So what does this all mean?
- 13:20Does this terminal growth rate act like the happily ever after
- 13:23fees of the model? Like after 10 years of running a
- 13:26marathon at a 15% Sprint, the company just suddenly settles
- 13:30into a permanent 2.5% jog for the rest of eternity.
- 13:33If we connect this to the bigger picture, that 2.5% terminal
- 13:37growth rate is one of the most vital anchors in macroeconomic
- 13:40modeling. It is not arbitrary at all, no.
- 13:44It represents the mathematical ceiling of maturity.
- 13:47Why 2.5 specifically though? Like, why wouldn't a highly
- 13:50successful company project a 5% or a 7% terminal growth rate
- 13:55forever? Because a 2.5% growth rate
- 13:59roughly aligns with long term macroeconomic inflation and the
- 14:03general growth rate of the global gross domestic product.
- 14:07Think about it. If an analyst assigned a
- 14:09terminal growth rate of, say, 8% to a single company forever,
- 14:13yeah, the math dictates that eventually that single company
- 14:16would outgrow the entire economy of the planet Earth.
- 14:19Oh wow. It would just consume the.
- 14:20Universe. It would consume the universe.
- 14:21A company mathematically has to cap out right?
- 14:24By setting the terminal rate at 2.5%, the analyst is modeling
- 14:28the transition from a hyper growth disruptor into a
- 14:30permanent stable fixture of the economy.
- 14:33OK, I see. The company will grow exactly as
- 14:35fast as the global economy itself grows, moving in tandem
- 14:38with inflation. So the company becomes so
- 14:40massive that it simply becomes part of the economic atmosphere.
- 14:43Yes, it stops fighting the wind and just drifts with the
- 14:46prevailing currents of global GDP.
- 14:48Beautifully put. So the 15% phase is the
- 14:51aggressive conquest of new territory, right?
- 14:53And the 2.5% phase is the Empire permanently maintaining its
- 14:57established borders. That is the exact mathematical
- 14:59life cycle of a successful corporate empire, perfectly
- 15:02translated into two variables in a Python script.
- 15:05That's wild. Yeah.
- 15:06So we have our deeply pessimistic stress tests and our
- 15:08empire building nominal timeline.
- 15:11The script runs all this profound timeline splitting
- 15:13math, and now we reach the final block of code where the script
- 15:18actually packages this prophecy. The output.
- 15:21The output and the methodology here is a stark contrast to how
- 15:24this kind of information is usually presented.
- 15:26Yes, it is. So the script takes the outputs
- 15:29and writes them to two highly specific markdown files, PDD
- 15:34financial report dot MD and PDD financial data scripts dot MD.
- 15:38Just raw text files. Just raw text files.
- 15:41Then he utilizes a Python function called OST dot walk to
- 15:44sweep through the directory gathering up these bare bones
- 15:48text documents and it compresses them using a method called zip
- 15:51deflated. The final output is just a
- 15:54single mundane file named PDD financial research package dot
- 15:58zip. You know that sequence
- 16:00generating raw text files and compressing them into a standard
- 16:03zip package? It represents a massive
- 16:04ideological shift in how financial research is
- 16:07distributed. Because honestly, it feels
- 16:10incredibly understated. I mean, we Mac out the ultimate
- 16:13fate of a global company projecting out to eternity to
- 16:16decide if it freezes in carbonite or becomes an economic
- 16:19empire and it all ends up stuffed into a mundane dot zip
- 16:22file. It's like writing the secrets of
- 16:24the universe on a post it note. Well, historically, Wall Street
- 16:27firms would packaged this exact same math into 100 page securely
- 16:32locked PDF. Yeah, they would hire graphic
- 16:34designers to build beautiful persuasive charts.
- 16:37The glossy charts. The entire goal was to make the
- 16:39presentation look as authoritative and impenetrable
- 16:42as possible. The aesthetic of the PDF is
- 16:45meant to say, you know, trust us, we did the complex math, you
- 16:48wouldn't understand it anyway. Precisely, the traditional Wall
- 16:51Street model was built on the concept of the black box.
- 16:54You received the glossy PDF, you read the analysts conclusions
- 16:58and you look at the beautiful. Charts.
- 17:00You can't touch it. You cannot easily see, and you
- 17:02certainly cannot alter the underlying assumptions.
- 17:06The presentation is designed to obscure the mechanics and
- 17:09elevate the authority of the analyst.
- 17:11Wow, so this zip file is like a rebellion against that.
- 17:14Exactly. A zip deflated package
- 17:17containing raw markdown files and Python scripts is the exact
- 17:20opposite, right? It is the democratization of
- 17:23financial modelling. Markdown is just raw,
- 17:26unformatted text. Anyone can read it?
- 17:28Anyone. The zip file doesn't just
- 17:31contain a static report, it contains the actual engine.
- 17:33Oh, I see. It allows any independent
- 17:36researcher anywhere in the world to download the package, extract
- 17:39the code, and look directly at that -.0909 assumption.
- 17:44And more importantly, they can change it.
- 17:45They can run their own stress tests, like if they disagree
- 17:48with the hurricane prediction of a 9.09% drop, they can change
- 17:52the variable to a 4% drop. Hit run and instantly watch how
- 17:56that single change cascades through the 10 year time machine
- 18:00and alters the final valuation. The output is no longer a
- 18:03finalized pronouncement handed down from an ivory tower.
- 18:07It is a portable laboratory. Portable Laboratory.
- 18:10I love that the OST dot walk function sweeping up the scripts
- 18:13and data is essentially packaging the evidence so the
- 18:15end user can cross examine the analyst's logic.
- 18:18Right. It shifts the power dynamic from
- 18:20the institution with the best graphic designers to the
- 18:22individual with the most robust logic.
- 18:25That completely reframes the entire final block of code for
- 18:28me. The zip file isn't an
- 18:29anticlimax, it is an invitation to elaborate and verify.
- 18:33Transparency is the point, yeah. I mean, we've unraveled a time
- 18:36machine, mapped out the stress test of a shrinking utility,
- 18:40modeled the conquest of a corporate empire, and packaged
- 18:42it all into an open source laboratory.
- 18:45It's remarkable how much philosophy and economic theory
- 18:47is embedded in just a few dozen lines of Python.
- 18:51It really is. By reading the code directly,
- 18:53you bypass the narrative entirely.
- 18:55You get straight to the mechanics of how the future is
- 18:57valued. You know, it is fascinating to
- 18:59see how ruthlessly the market mathematically discounts the
- 19:04value of the future. Very ruthless.
- 19:06Like the script demands a 10% tax on tomorrow simply because
- 19:09tomorrow is uncertain, right? And it makes you wonder if we
- 19:13inadvertently do the exact same thing in our own lives, just
- 19:17without a Python script warning us about the math.
- 19:20Are we applying a steep discount rate to our own futures?
- 19:23Exactly. Applying financial mechanics to
- 19:25human behavior is always a revealing exercise.
- 19:28The parallels are definitely there.
- 19:29Think about the daily choices I make.
- 19:32Do we prioritize immediate comfort today?
- 19:35The dollar in our pocket right now.
- 19:36Right over our own long term growth and potential ten years
- 19:39down the line? Like, are we heavily discounting
- 19:42the value of our future health, our future skills, or our future
- 19:45happiness simply because those things feel too far away and too
- 19:49uncertain to invest heavy effort into right now?
- 19:52Are we taxing our own future just to prioritize the present?
- 19:55Exactly. It really makes you evaluate
- 19:58whether your own personal trajectory is in a 15%
- 20:01compounding Sprint or if you have quietly accepted a 0%
- 20:06terminal growth rate. If you run your own personal
- 20:08choices through a discounted cash flow model, the results
- 20:11might be far more confronting than any corporate valuation.
- 20:15Definitely something to Mull over while you evaluate your own
- 20:18daily routines. And as a final reminder, the
- 20:20Python scripts and financial scenarios we explore today are
- 20:24based on publicly available data, strictly for research and
- 20:27entertainment purposes and do not constitute financial advice.
- 20:31Always consult A lysis professional.
- 20:34And conduct your own due diligence.
- 20:36And if you want to dive deeper into market tracking, don't
- 20:38forget to check out the Buffett Indicator or Bubble Watch apps
- 20:41or the Deep Values app for advanced agent research.
- 20:44Link in the description below. Understand the architecture of
- 20:47these models is the best way to demystify the future.
- 20:50Keep digging beneath the surface and we will see you on the next
- 20:54deep dive.