Last night, you probably saw three headlines and thought they were three separate stories.
The Dow fell 561 points.
Gold fell below $3,800, hitting a one-year low.
Nvidia beat earnings expectations, but its shares closed down 2.89%.
Three stories.
Actually, it was one move: on the same day, $81.3 billion shifted from the secondary market to the primary market.
This morning, Anthropic listed on the Nasdaq under the ticker ANT.
Priced at $83, above the top of the roadshow range.
After increasing the number of shares offered, it raised $81.3 billion.
Plus a concurrent $2.6 billion private placement, with Nvidia, Google, Microsoft, Micron, AMD, TSMC, and Broadcom all on the list.
This is the largest IPO in U.S. history, surpassing the SpaceX offering in June.
And the liquidity drain didn’t start today.
1. The liquidity wasn’t drained today; it happened during those two weeks of the roadshow
Money for IPO subscriptions has to be locked up in advance.
For an offering worth a trillion dollars, the real liquidity drain is the cash locked up between the public prospectus and pricing, along with the holdings sold to raise that cash.
This is documented in black and white.
In a CNBC article published Monday about Bitcoin falling below 62,000, it listed “Anthropic’s IPO pulling speculative money out of risk assets” as one source of pressure.
Look at the timeline.
Bitcoin was still around 74,000 on November 11.
It broke below 70,000 on the 13th, then plunged to 62,000 on the night of the 15th.
It closed at 62,984 last night, marking six consecutive weeks of declines—the longest losing streak since February.
Over the same period, gold fell from 4,139 on November 9 to 3,727 last night.
These aren’t two markets making the same mistake at the same time.
It’s the water levels in two pools being lowered by the same pipe.
The money hasn’t disappeared. It’s in this morning’s subscription queue.
2. The Dow falls 561 points while the Nasdaq rises: the index is deceiving you
The closing figures for the three indexes last night looked awkward side by side.
Dow: 50,198.17, down 561 points, or 1.11%.
S&P 500: 7,323.63, down 0.23%—its fifth consecutive decline and longest losing streak since June.
Nasdaq: 24,429.88, up 0.42%.
Same night, one fell and one rose—a difference of more than a percentage point.
What you need to look at is what happened under the surface.
Of the Dow’s 561-point decline, UnitedHealth alone accounted for about 236 points.
It fell 10.21%.
The reason was a proposal CMS put forward on Monday to expand audits of Medicare Advantage and claw back historical overpayments—about $90.2 billion over ten years.
Humana fell 13.04% that day.
Of the 30 Dow components, 20 closed lower.
Now let’s look at the Nasdaq side.
On the Nasdaq exchange, 1,769 stocks fell and 1,463 rose that day.
The index gained 0.42%, even though more stocks fell than rose.
Why? Because the weight is concentrated in just a handful of stocks.
This is the first thing I want you to remember today: an index is a weighted average; your portfolio is equal-weighted.
If you own an index, you made 0.42% last night.
If you own a basket of stocks, chances are you lost money last night.
Same market table, two different experiences. The difference is a weighting table.
3. Good news is now only good for 12 hours
Nvidia reported fiscal third-quarter revenue of $115.8 billion and guided to $130 billion for the fourth quarter. Its fiscal-year growth target was raised from 70% to 75%.
It rose about 3% before the market opened.
It closed down 2.89%.
An earnings report that beat expectations went from up 3% before the open to down 2.89% at the close.
What happened in between?
At 2 p.m. Eastern, the minutes from the October meeting came out.
“Almost all participants” thought further rate hikes “would likely be appropriate” to bring inflation back to 2% on time.
“A few” said that if inflation data were poor, a move larger than 25 basis points “could be considered” in December.
“Some” had already argued for a 25-basis-point hike—not just the two who voted against the decision.
The dissenters were Cook and Hammack. Both favored an immediate 25-basis-point hike.
Not one person in the entire set of minutes discussed rate cuts.
That morning, Waller said on CNBC that a half-point hike should be considered in December.
The market now puts the odds of a 25-basis-point hike in December at about 95%; the odds of a half-point hike are in the single digits.
So the earnings report wasn’t useless.
It just didn’t have pricing power.
Pricing power lies with the December meeting, and the minutes have already made the direction clear: the question isn’t whether to hike, but by how much.
A $115.8 billion quarter can’t outweigh the words “almost all participants.”
Good news now has a shelf life of just 12 hours, because the direction of discount rates hasn’t changed.
The premarket rise was retail buying. The 2.89% drop at the close was institutions settling up.
4. Gold isn’t falling because of inflation; it’s the last pillar of buying that’s giving way
Here’s a detail many people might miss.
Last night, the 10-year Treasury yield was 5.07%, down 3 basis points from the day before.
Yields are falling.
Gold still fell 3.17%, dropping to 3,742 intraday—its lowest level since September 2025.
If gold were simply moving against interest rates, it should have risen last night.
It didn’t.
That tells us the marginal sellers in this leg down aren’t interest-rate traders.
So who’s selling?
Look at the structure of demand.
In the third quarter, global physically backed gold ETFs saw net inflows of 211 tonnes, and holdings hit a record.
Central banks bought 675 tonnes in the first three quarters, more than in the same period last year.
That sounds pretty solid.
But the picture started to diverge in October. Global gold ETFs are still seeing net inflows, while North America saw net outflows of $512 million.
And the biggest official buyer, Poland, has already reached its 700-tonne target.
Third-quarter bar-and-coin demand was down 19% year over year.
Third-quarter jewelry demand was down 17% year over year—the weakest third quarter in this data series.
In other words, the two pillars supporting gold prices are still there, but the third—retail and bar-and-coin demand—has already pulled back.
Pricing power now lies with ETFs.
ETFs are the only money that makes decisions based on what it sees in the market.
Once they turn, there’s no one left to catch the fall.
From its January peak of 5,608, gold is now at 3,727—a 33% drawdown.
This is its worst year since 2015.
Last year, you bought gold for “inflation” and “safe-haven protection.”
Inflation is running hot this year, and gold has fallen by a third.
That’s because gold has never been priced on inflation, but on real interest rates. The 10-year real yield is 2.88%, and the dollar is at 103.81, a 19-month high.
The logic is straightforward: high inflation leads central banks to raise rates; rate hikes push nominal rates above 5%, and real rates rise along with them.
Gold pays no interest. Every step up in real rates increases the cost of holding it.
So the idea that “inflation means buy gold” only holds when central banks aren’t raising rates.
Central banks are raising rates this year, so that idea doesn’t hold.
The reverse is true, too. Once central banks pivot, gold doesn’t need any new bad news; it only needs existing buyers to stop adding to their positions.
5. The order of selling is your risk map
Put last night’s three events in order, and you’ll see a queue.
First in line: sell assets with no cash flow. Gold -3.17%, silver -2.65%.
Second: sell assets with policy risk. UnitedHealth -10.21%, Humana -13.04%.
Third, sell assets that generate cash flow but are expensive: Nvidia -2.89%.
Fourth: assets that were already falling. Bitcoin at 62,894, down 0.14% over 24 hours, with a low of 62,000.
This queue isn’t ordered by what’s “good” or “bad.”
It’s ordered by whose money has the shortest staying power.
Insurance companies’ money depends on policy, gold’s on real rates, Nvidia’s on next year’s capex, and Bitcoin’s on leverage.
Four pools, four different maturity dates.
When the water level drops by one notch, the shallowest pool dries up first.
This isn’t a prediction; it’s an ordering of what has already happened.
So what you think of as diversification is often just putting the same money into four containers with different maturity dates.
You diversify by asset; the market looks at maturity dates.
When the water level shifts, all four make a sound.
6. Watch just three numbers today
First: ANT’s debut.
If it jumps at the open while the Nasdaq and Bitcoin don’t move, then the $81.3 billion came from existing funds—the liquidity drain thesis holds.
If ANT rises and the Nasdaq and Bitcoin rise too, that means new money is coming in, and this article’s thesis should be discounted.
Watch both directions. Don’t look only at ANT.
Let me be clear: the IPO liquidity drain and the declines of the past few days coincided in time. I can’t prove it was the only cause.
So don’t treat it as a conclusion. Treat it as a hypothesis that can be tested today.
Second: after gold’s drop to 3,727, does it hold 3,700?
At the same time, watch whether the real yield at 2.88% moves above 3.00%, and whether the dollar at 103.81 gets above 104.
If real rates rise above 3.00%, 3,700 will just be a number gold passed on the way down.
Third: Bitcoin at 62,000.
That was last Saturday’s low, and it’s also the level Glassnode has marked: if the daily close falls below 62,000, the next level to watch is 53,700.
The first resistance above is 65,000. Until Bitcoin gets back above it, any rebound is just a recovery.
It’s now at 62,894, just over 1% above 62,000.
Two reference points for you: the 20-day moving average is at 72,826, and the 200-week moving average is around 67,800.
With the price below both lines, this is a downtrend structure, not a pullback.
7. Three things not to do today
First, don’t chase ANT on its debut.
The debut of the biggest IPO in history is for underwriters and original shareholders, not for you.
If you want to participate, wait for it to return to around its offering price, or wait for the first earnings report.
Second, don’t add leverage above 62,000.
If you want to go long, wait until the daily close is above 65,000. Don’t borrow a ladder at the edge of a cliff.
Third, don’t try to call a bottom in gold at 3,700.
This year’s lesson from gold isn’t “it’s fallen too far”; it’s “it can keep falling longer than you think.”
UBS says this is a positioning adjustment, not a structural problem, and sees gold at 4,500 by year-end.
ECR Research says to look for 3,500 first.
Both are guesses. Don’t guess either.
Waiting for real rates to turn is more reliable than waiting for a round-number level.
Last night’s 561 points, $81.3 billion, and 3.17% weren’t accidents.
They were the end of a settlement process that had already been lined up.
Your job isn’t to guess the next number; it’s to figure out where you stand in the queue.
(Data as of 2026-11-19 08:00 CST; U.S. stocks at the November 18 close, crypto from live Kraken spot prices, and commodities based on November 18 intraday and closing prices.)
(This article is market analysis and does not constitute investment advice. Data comes from public sources.)
— MK守约 #守约交易哲学

