Big news! US stocks surged 16% in two months, hiding the same bomb as 1987 Black Monday: 63% zero-day options are ticking, AI bull market not as stable as you think!
16% in Two Months
I repeat
Not annualized
Not two years
It's this April and May
In just two months
S&P 500 up 16%
How many times has this happened in history?
Counting this time, 5 times. From 1928 to now, nearly 100 years, only 5 times.
Charles Schwab Chief Investment Strategist Liz Ann Sonders said the other day:
Over the past two months, only 17% of S&P 500 components outperformed the index itself.
17%! So this 16% gain is being pulled by a handful of stocks; the other 83% are either lagging or dragging.
You don't know whether to be happy or worried.
Happy that your account may be rising.
Worried that if so few stocks are supporting the market, what happens if those stocks falter?
Frankly, this isn't a question of whether something will happen.
It's: When it does, how bad will it be?
Hello everyone, in this episode I want to discuss a topic that chills me the more I research—behind the 16% US stock surge in two months, three dark mines are simultaneously lit.
Each mine alone is not fatal, but all three lit together with an invisible connection is the deadliest part.
And there's a structural issue almost identical to October 19, 1987, the most terrifying day in US stock history.
Not the trend, but the structure—the pipes, leverage, and "what-ifs" that could bring down the whole system are the same as 1987.
So this episode will break it down in three steps:
-
Who is buying this 16% surge, what tools they use, and consequences.
-
Where does the liquidity you think is there come from, and when will it run out.
-
The core narrative of the rally—AI—is about to hit a physical wall.
Three topics seem unrelated, but together tell the same story.
First Mine: Microstructure—Zero-Day Options and Gamma Bomb
1987's Ghost: Portfolio Insurance
October 19, 1987, Black Monday, Dow fell 22.6% in one day.
If you were there, you'd see the market open down, accelerate, and fall faster. Traders couldn't hear each other over the noise because everyone wanted to sell and no one bought.
22.6% in today's market would mean about $9 trillion evaporated in a single day.
The culprit was not bad news, war, or economic data, but something called "portfolio insurance."
Simple principle: institutions set up stop-loss programs in computers to automatically sell futures when the market fell 3%. But everyone set similar thresholds, creating cascading selling that amplified the crash.
The 1987 flash crash wasn't a reaction to bad news, but a defect in market structure that ignited itself.
Today's Structural Twin: Zero-Day Options
Now guess if there's something in today's market with the same logic as portfolio insurance but dozens of times larger?
Yes. And it's grown beyond imagination.
It's called zero-day options (0DTE): bought today, expire at today's close. Pure intraday gambling.
-
Three years ago, 0DTE was less than 10% of S&P 500 options trading.
-
By end of last year, it rose to 45%.
-
Just last month (May 20), CBOE reported 0DTE now accounts for 63% of S&P 500 options volume.
Every day, over 2/3 of options bets are on direction before the close.
You might say: It's just a casino; someone loses, someone wins, so what?
The problem isn't the gamblers; it's the market makers.
Market Maker Gamma Reversal: From Stabilizer to Amplifier
When you buy an option, someone sells it—market makers. They hold large portfolios of options, and price changes create gamma, which dictates how much they must buy or sell to hedge.
-
If market makers are long gamma, they buy when market drops, sell when it rises—like a spring, dampening volatility. Good.
-
If they are short gamma, it's reversed: buy when up, sell when down—they become accelerators, pushing moves further.
When the market is calm, positions are comfortable and gamma is positive. But after a continuous fast rise (like 16% in two months), dealer gamma flips from long to short. Once flipped, the stabilizer becomes an amplifier.
In April, a research report estimated current market maker gamma exposure at roughly negative $7.5 billion.
Meaning: if S&P drops 1%, forced gamma hedging could add another 1%—turning 1+1 into 2.
A 2% daily drop might not seem much, but remember 0DTE expires in hours. Over 2 million contracts trade daily. Gamma explodes exponentially last two hours. If any bad news hits an hour before close, dealer selling is not linear—it's exponential.
This has been warned by BIS, SEC, and even CBOE, but market ignores while indexes keep hitting highs.
Second Signal: Short Volatility
Next signal: short volatility.
VIX (fear index) is near 15 after 16% rally. Yesterday (June 3) closed at 16.2, near one-year low.
But look at positioning: August 2025 data shows net short VIX futures surged to 92,800 contracts, near all-time high.
Shorting VIX means betting volatility stays low—essentially betting no surprises—no Iran aggression, no yield spike, everything runs smoothly.
This position is like driving 180 mph on the highway without a seatbelt, telling the passenger "no curves ahead."
Are there really no curves ahead?
Second Mine: Macro Liquidity—Reservoir Dry and ERP at Record Low
Secret of Shadow Liquidity: US Treasury
How did US stocks rise 16% in two months while Fed kept rates high? Where did liquidity come from?
Answer: US Treasury.
During COVID, Fed injected massive liquidity via QE. Much of it sat in Fed's reverse repo (RRP) facility, peaking at $2.5 trillion at end of 2022. Over past two years, this reservoir drained as Treasury issued T-bills, drawing money from RRP into the banking system and into buying power.
Wall Street calls this "shadow liquidity"—not from Fed, but indirectly from Treasury issuance.
The Pump Stopped
By June, the reservoir is nearly empty. Fed's overnight RRP balance is only about $2 billion, essentially zero from the $2.5 trillion peak.
The invisible pump that quietly fueled stocks is shut off. Any extra liquidity pressure will now hit bank reserves directly. Tightening could affect repo markets and hedge fund funding.
Remember Sept 2019: repo rates spiked from 2% to 10% in a day due to a liquidity bottleneck. That was just a preview. Today's pipeline is more complex.
Key: long-end Treasury yields are creeping up. Yesterday 10-year yield closed at 4.49, up ~30 bps since April while stocks rose 16%—bond yields rising, stocks diverging.
This was also seen in August 1987, two months before Black Monday. Yields jumped from 7% to nearly 10%, S&P kept rising, then collapsed.
History doesn't repeat exactly, but markets sometimes ignore a problem until forced to face it.
Equity Risk Premium: Historically Low Compensation
Why is the market ignoring rising yields? Because one metric makes stocks look cheap: equity risk premium (ERP).
NYU professor Damodaran calculated S&P implied ERP at 4.23% in January, near bottom of past 20 years (average 4.5-5%).
Translation: you're getting historically low risk compensation for stocks. Since then, S&P rose another ~10%, so ERP has likely broken below 4%—the market is paying the highest price for risk.
I'm not predicting a crash. I'm just telling you: every dollar of gain you earn right now comes with historically rare low compensation. If that's not enough to make you cautious, the third mine is the most concrete.
Third Mine: Physical Constraints—AI Hits Grid Ceiling
Who makes the world's largest transformers? China, South Korea, Germany, not the US.
How long to deliver a large HV transformer? 2 years if ordered 3 years ago; 5 years if ordered today.
S&P Global article on May 9 tracked 140 AI data center projects announced in US for 2026, total planned capacity 12 GW (like NYC's electricity). But only 5 GW actually under construction; 11 GW still "announced" with no foundation dug. A data center takes 12-18 months from digging to operation.
Why Not Start Construction?
Not lack of money—tech giants have doubled AI capex estimates this year. The bottleneck is grid and transformers. A large AI data center needs hundreds of MW, requiring large HV transformers. Global manufacturers (Siemens, Hitachi, GE) have 5-year lead times. GE's executives said they've sold much of 2029-2030 capacity, leaving only ~10+ GW slack globally by 2030. Compare to just US announced 12 GW.
Government Emergency Authorization: Cut AI Power First
In May, PJM (covering 13 states) got US Dept of Energy emergency authorization to directly cut data center power in emergencies.
Think about it: government gave grid operators the right to shut off AI first—not homes, not hospitals, but data centers.
It's dark humor: Wall Street touts AI as the next industrial revolution, but a large portion of chips leaving factories have no powered racks waiting. They arrive at dark data centers.
Morgan Stanley in February: Developers expect grid bottlenecks will delay a significant portion of announced data center projects by 2027-2028.
Meanwhile, US utility capex forecast for 2026-2030 increased to $1.3 trillion (up $300 billion in one year), all for physical stuff: lines, substations, transformers—no code, no AI, just iron, copper, and steel.
Triple Mismatch
- Money mismatch: Wall Street valuations for AI assume capex can land, data centers built, compute monetized. If grid prevents construction, invested money becomes sunk cost. GPUs sitting in powerless warehouses depreciate every quarter.
- Time mismatch: AI data centers move at internet speed (2 years from site selection to operation), but grid moves at industrial age speed (new transmission lines take 7+ years).
- Attention mismatch: Everyone focuses on AI's 'intelligence' but ignores its 'body'—transformers, power lines, cooling water.
Already Happening
May 19 Reuters: PJM already has authority to limit data center power. May 9 S&P Global: half of US 2026 AI data center plans effectively shelved due to transformer shortages.
Three Mines Simultaneously Lit: Market Prices a Perfect World
- First mine (microstructure): 0DTE options push gamma exposure to unprecedented levels, turning market makers from stabilizers to amplifiers. Any small dip could trigger an avalanche.
- Second mine (macro liquidity): Fed RRP reservoir empty, shadow liquidity gone. Long yields rising, ERP at 20-year low—more risk, less compensation.
- Third mine (physical constraints): Core narrative AI hits real wall: transformers, grid, generators insufficient. Not a money problem but time, capacity, physical limits.
These three mines belong to different domains but are connected: market prices a perfect world—no accidents, infinite liquidity, unlimited AI expansion. But in the real world, gamma amplifies surprises, liquidity pump is off, AI hits physical ceiling.
In summer before 1987 Black Monday, the market also priced a perfect world. Then structural products laid mines, rising rates were ignored, everyone thought 'this time is different.'
The difference? Today's system is more complex, with more layers of leverage and interlocking gears. Back then only futures and stocks malfunctioned; now add options market makers, passive funds, CTA trend-following funds, vol control funds, risk parity funds—five players each with forced mechanical rules. One trigger can knock them down like dominoes.
CBOE on May 20 said 0DTE options now 63% of S&P 500 options volume. They used the word 'Remarkable'—I read it as ominous.
Imagine a building guard says 63% of daily visitors have passes issued today that expire tonight. Is the security system reliable?
What Should Ordinary People Do?
I won't give buy/sell advice. Just want you to know: when your account is up and everything looks rosy, three things are ticking in the dark.
You don't have to run, but you should know where the exit stairs are.
One quote from a trader after 1987: 'Before the crash, you think the market is deep and can withstand anything. After the crash, you realize its depth is just a thin layer of glass—underneath it, nothing.'
I'm not predicting a crash tomorrow. But when you're standing on that glass, it's better to know it's glass.
That's all for this episode. Like, share, subscribe if you found it helpful. One question: which of the three mines do you think is most dangerous?
