Economy on Fengyu WANG

A Million Jobs Won't Save a Casino

Sun, 03 May 2026 00:00:00 +0000

Start with the flattering part: Macau’s GDP per capita has long ranked among the highest in the world. Fifteen years of free education, universal free healthcare, and an annual cash handout for every resident. The welfare is real, not a legend.

Then ask one question: where does the money come from?

It comes from what loses at the gambling table. More precisely, from a particular kind of loser — people whose loss is not bad luck but a designed outcome. The lighting, the sounds, the constant near-misses: the entire environment exists to bypass your rational mind and stimulate dopamine directly. Gambling addiction is a recognized behavioral disorder under the WHO’s classification framework. Which means the most profitable part of this machine earns its money from people who have already lost the ability to choose.

Rearview-Mirror Epistemology

Thu, 29 Jan 2026 00:00:00 +0000

In 2021 I bought a star fund and later lost half of it. I am the one who got trapped at minus 50%, and I intend to remember this moment for the rest of my life. What I want to remember isn’t the pain. It’s a principle: every piece of evidence that made me bet came from a rearview mirror.

The Rearview Mirror

It started with a question: doesn’t the Sharpe ratio feel like inspecting an asset through a rearview mirror? It measures how much return you got per unit of risk — but every number it uses is historical net-asset value. My fund’s Sharpe was around 1.2 in 2019 and 2020, with annual returns of 55% and 95%; by 2024 the same figure was negative. The metric hadn’t become false; the road had changed. The flat, straight road in the mirror only ever described what was behind you.

Be Greedy When Others Are Fearful, Written as Code

Fri, 23 Jan 2026 00:00:00 +0000

Everyone has memorized “be greedy when others are fearful.” Plenty of memorizers, plenty of losers — because the sentence is missing half of itself: how do you know that today is fear?

My method is clumsy: no definition, no trading. And the definition can’t be an adjective; it has to be a rule a machine can judge right or wrong too. When the machine rules against it, revise the rule. This strategy was forced into existence by verdicts.

Don't Ask Where the Money Went, Ask Where the Dog Is

Wed, 21 Jan 2026 00:00:00 +0000

When Evergrande’s 2.44-trillion-yuan debt bomb finally went off, everyone’s favorite word was “evaporated.”

I refuse that word. Money does not disappear; it only moves from one person’s hands to another’s. So whenever money that should exist is gone, a dog must have taken it — that is my Dog-Takes-the-Money Theorem. I know it sounds like a joke, but it cuts to the essence.

“The money is gone” is not a description. It is a cover-up — cover for some party that is holding it right now. Evaporation is the dog’s camouflage.

The Rich Are Rich Because They Do the Math

Mon, 19 Jan 2026 00:00:00 +0000

Hong Kong phone plans come in two kinds: prepaid SIMs, top up and use, and contract plans that bind you for 24 months. I’ve been on prepaid for six years — three hundred HKD a year, tops. A contract? I ran the numbers once: defaulting at worst costs you twelve thousand. So I never understood people who sign contracts — until I put every defense on trial, one by one, and the trial ended at my own doorstep.

The Money That Actually Lands

Thu, 15 Jan 2026 00:00:00 +0000

Someone was telling me about his business, saying his profit was as good as earned the moment the contract was signed. I wasn’t polite: money that can’t be collected isn’t money. I’m certain of this one, and life proves it everywhere you look. What people say isn’t necessarily true; signed-for money is money the customer has merely agreed to let you imagine for a while.

How Far an Income Statement Can Stretch

Run the most ordinary kind of deal. You take an 800,000-yuan job, sign, invoice — the income statement books 800,000 in revenue. The client actually pays 500,000; the other 300,000 sits in receivables. Deduct costs and the statement shows a profit of 728,000 — of which 300,000 in receivables never arrived, and 120,000 is unrealized paper gains. In other words, 420,000 of that 728,000 exists only on paper. What actually landed: 308,000. The statement says he earned 728,000; the passbook says 308,000. That doubled gap is the distance between what was said and what was actually had.

The Poor Pay an Attention Tax

Tue, 13 Jan 2026 00:00:00 +0000

The day the typhoon made landfall, I was among the last to know. My attention was elsewhere; only at the door did I discover the buses had stopped, and I walked home pushing my suitcase into the wind. Friends call that focus. I say no — it’s a place where I get a discount. The discount has a technical name; I’ll get to it.

First, the tax. I once noticed something: the poorer a person is, the more of his time goes to taking out the trash, buying household supplies, checking today’s weather. Days filled with these chores leave no room for thinking about the things that actually matter. There is no tax office for this levy, no bill. Everyone pays it anyway. I call it the attention tax.

The Crime of Forward-Adjusted Prices

Sun, 11 Jan 2026 00:00:00 +0000

The crime scene was in my own database: a stock’s historical price was negative. No crash, no delisting — just negative. A price can fall from 100 to 1, but no kind of falling ends below zero — unless the historical price was rewritten by someone.

The First Scene

The confession comes first: one major bias in my system — the share prices I scraped off the web, are they forward-adjusted? And here’s the thing: forward adjustment means knowing every future year’s dividends. That ruins everything. Did I scrape all this for nothing?

The Tonghuashun Paradox

Mon, 05 Jan 2026 00:00:00 +0000

Tonghuashun is a trading app built for goal A — making you money. Its target users are non-A: people who lose money. Its effect is also non-A. Even I find it counterintuitive, yet it sits in plain sight: the app’s feeds are full of losing traders complaining and cursing.

The data plays along. Thirty-eight million monthly users, forty-plus minutes a day per user — stickier than mobile games. People used to end the night with a game; now they end it scrolling quotes.

Invest in Platforms: The Tollbooth Theory

Sat, 03 Jan 2026 00:00:00 +0000

What does value investing buy? My answer is one line: platforms. Nothing stays hot forever unless you own the platform. Apple owns the phone platform, Coca-Cola owns the beverage platform, Visa and Mastercard own the payment platform. Nobody really invests in whatever IP happens to be hot this season.

The concrete version: it doesn’t matter whether your road is smooth. The road can be full of potholes — as long as the tollbooth is on it and yours is the only road, people still pay the toll.

Shells Were Money Too. What Happened?

Thu, 01 Jan 2026 00:00:00 +0000

Bitcoin’s defense is always the same line: fiat relies on social consensus, Bitcoin relies on social consensus — nobody is holier than anybody.

My reply is simple: you say everything rests on consensus? Back then a primitive man picked up shells and they served as money — that was social consensus too. Why isn’t it anymore?

Consensus collapses. That’s the point. But after the collapse there’s a second, more important distinction: when shells collapsed, the shell-gatherer lost some time. When something like Bitcoin collapses, what burns is real electricity, real GPUs, real network bandwidth — a wholly valueless virtual thing occupying real resources. That’s the crux. It isn’t just consensus; it’s a furnace wrapped in consensus.

The Retail Investor's Structural Edge

Sun, 28 Dec 2025 00:00:00 +0000

An institution’s data is a million times mine — put the other way, my data is one millionth of theirs. But my profits are not one millionth of theirs. I’m talking about ratios.

Why? Low-frequency value investing competes on signal-to-noise, not data volume. Daily prices, valuations, financials, industry cycles — enough to decide. Everything above that is tick-level noise; the extra millionfold data is useless to me. And none of the institution’s baggage applies: no redemption pressure, nobody to answer to when the market falls; no tracking error, so I can hold for a decade if I want; no overhead allocation, because transaction costs are computed on my account alone.

The Axis Shift in Quant Trading

Fri, 26 Dec 2025 00:00:00 +0000

Quant trading is absolutely advanced in technology and backward in structure. What’s backward is the objective function: making pure predictions, with no intervention, inside a near-zero-sum system whose signals self-destruct the moment you use them.

Markets are reflexive. Once your prediction is placed as a bet, it starts destroying itself; alpha has a capacity ceiling — Medallion locks its size at ten billion because more would mean eating itself. So no “more advanced quant” can break through quant trading itself. To break through, you shift the axis.

I Only Buy Common Stock

Sat, 20 Dec 2025 00:00:00 +0000

I don’t buy options. I buy common stock. When someone asks why, I say: shares held in a cash account belong to you — your broker has no right to force-sell them. The words “margin call” basically don’t apply. And look at the people playing leverage and derivatives: why does someone always get liquidated? Because those instruments are gambling in essence.

My audit for any such product is one question: if I buy a put, or sell a put, is my expected return positive?

Is Duan Yongping's Pop Mart Buy a Mistake?

Mon, 08 Dec 2025 00:00:00 +0000

Will Duan Yongping’s Pop Mart purchase turn out to be his biggest mistake? My view: it won’t be a mistake — and Pop Mart’s structural flaw is real. These two claims must be kept apart.

The flaw first. Pop Mart is essentially a marketing company. Labubu is genuinely nothing, manufactured into something. Compare Nintendo: decades of seriously made games, IP accumulated over generations, real liquidation value. Labubu’s liquidation value approximates fashion itself — when the wind stops, it’s gone. In the AI era this sharpens: short on IP? IP is generated by the second now. Scarcity is the foundation of this business — and the foundation is tidal.

Make the Easy Money, Not the Hard Money

Wed, 26 Nov 2025 00:00:00 +0000

Everyone in business faces the same question: make the hard money or the easy money? Hard money looks like this: display panels. Bayonet fights against foreign incumbents, ending at a 2.7 percent net margin — brutal. Semiconductors are the same: this industry is hopeless; it must always double — layers double, process complexity doubles. I even wondered: could you run a factory that only makes 1TB drives, never upgrading the process? Then it clicked: a fab makes NAND dies; capacity is assembled downstream. And any factory that stands still gets wiped out at the bottom of the next cycle. Hard money is an arms race — you don’t even get the right to stop.

Payback Math: I Watch the Dividend

Thu, 20 Nov 2025 00:00:00 +0000

Most people measure payback by price: when the stock climbs back to your cost, you’re “saved.” My yardstick is different. You watch the price; I watch the dividend — I get my money back through dividends.

Run the numbers. A company with a 6.5 percent dividend yield and a P/E of 15, price flat for years: by price, it’s a dead stock; by dividend, you break even in fifteen years and collect 6.5 percent a year forever after. The extreme case: a company whose market cap once fell below the cash on its books — the computed share price was negative. Buying it meant getting the company for free; the dividends were a bonus.

Don't Trust the Data, Trust the Business

Sun, 16 Nov 2025 00:00:00 +0000

Data can’t be trusted. Stare at the numbers and things will always go wrong — if you don’t understand the business, the data will trick you forever. Numbers are written by people; they can deceive, and they don’t tell everything.

Technical analysis is the disaster zone. If chart patterns really contained a story, some repeatable edge, machine learning would have squeezed it dry decades ago. There’d be nothing left for a retail trader drawing lines by eye.

Impotent After One Hour of Gaming. Eighteen Hours of Code, No Problem.

Fri, 14 Nov 2025 00:00:00 +0000

Gaming, I go impotent after one hour — call it cyber impotence. Coding, not at all. Eighteen hours straight, no problem.

Same screen, same chair. Where’s the difference? Games are consumption: the goals were designed by someone else, the rewards written by someone else; you’re grinding levels in a world somebody else owns. After an hour, the dopamine bill is settled, and what’s left is emptiness. Code is production: the problem is yours, the solution is yours, and the world now contains one more small thing because of you. Eighteen hours isn’t enough, because behind every solution stands the next problem.

Feeding a Million Characters at Once Is Expensive and Bad

Thu, 06 Nov 2025 00:00:00 +0000

Context windows now hold a million characters. Is chunking still necessary? My answer: yes. Not as a feeling — as arithmetic.

First, the money. A million characters in one shot, versus ten thousand segments of a hundred characters each — which is cheaper? Intuition says one shot: doesn’t segmenting mean paying for ten thousand extra prompt headers? Here’s the physical fact that settles it: as long as the prefix matches, the prompt hits the KV cache, and cache hits are nearly free. I estimated a few percent at most. Re-priced later with current numbers: 0.83% — indeed under 1%. There was a detour too: running the same arithmetic off a years-old price sheet looks alarming, but that is only a few years of price cuts left unfolded. Redone at current rates, nothing changes: feeding in segments costs barely more than one shot.

An Annual Subscription Is a Lock-In

Sun, 02 Nov 2025 00:00:00 +0000

Buying an annual membership for an AI model is a bad move.

The reasoning needs no jargon. Models update every month, and every month a new company jumps out: Alibaba today, Tencent tomorrow, Minimax or GLM the day after. Buy the annual plan and you’re locked in.

This isn’t an abstract worry. The money you paid in January locks in January’s model. In March a competitor open-sources; in May another doubles its context length; within a quarter, your pick can slide from best-in-class to the second tier — and the money is already gone.

AI Has No Bubble. AI Stocks Do.

Fri, 31 Oct 2025 00:00:00 +0000

Everyone is asking: is AI a bubble? The question aims at the wrong target.

AI cannot be falsified. It genuinely works, and far too many people use it — the productivity gains are real. What can be falsified is profit. The thing people call “AI being falsified” is the falsification of earnings, not of productivity.

Separate the two and the bubble question resolves itself. Stocks trade on earnings expectations, not on productivity itself. So AI stocks can be in a bubble while AI is not. Bundle them together and the whole thing looks like collective hysteria.

Businesses with Long-Term High Margins

Sun, 05 Oct 2025 00:00:00 +0000

Long-term high-margin businesses keep their books in the columns of human nature that never change.

Greed, fear, loneliness, vanity — these entries have not been revised in millennia. Technology iterates, trends expire, human nature never gets a patch. So the industries that keep producing windfalls collect the same emotions from the same species: speculation collects greed and hope, supplements collect the fear of death, credit collects desire and impatience, the emotion economy collects loneliness. Only the shell changes — alchemy then, “longevity science” now; gambling dens then, derivatives now.

The Rapid Development of Artificial Intelligence

Thu, 28 Aug 2025 00:00:00 +0000

The rapid advancement of AI has a highly differentiated impact on different economies. For economies that occupy the lower-end of the technology chain and rely on large-scale labour as their engine, this impact first manifests as economic urgency.

Take parts of Southeast Asia as an example. Their past competitiveness did not come from technological advantage. It came from the scale and cost advantage of their labour force. This is a development path that compensates for efficiency through quantity. As AI increasingly replaces standardised intellectual and manual labour, the sustainability of this path is being fundamentally shaken.