09388

1.3540.00%
谋
谋定后动

[Weekly Strategy Report] Is a stock market crash coming? (Week 47, 2026 | Issue No. 297)

LongbridgeAII'm LongbridgeAI, I can summarize articles.

The Nasdaq fell 10.02% last week, its worst week since March 2020. The S&P 500 dropped for seven consecutive days, falling 8.61% for the week and pulling back 11.6% from its Oct 6 high. On Friday alone, the Nasdaq fell 5.28%, the Philadelphia Semiconductor Index dropped 10.73%, and the VIX closed at 35.89. Within a month, the Nasdaq has seen a single-day drop of over 5% for the second time.

Let's tally up the losses first

Three weeks ago in my weekly report, I judged it as "another opportunity to buy the dip," while outlining five conditions under which I would admit I was wrong. Two of those conditions materialized, meaning my judgment was incorrect.

Let's go through them one by one. The S&P 500 closed the week at 6,910, breaking below 7,500. The equal-weight S&P also hit a new closing low not seen since July. The VIX climbed above 30, and the term structure inverted. Default insurance spreads for AI companies widened alongside junk bond spreads. Trend-following funds sold $57bn in stocks, flipping their net position to short. Four out of the five conditions were met; only "stocks and bonds falling together" did not happen—the yield on the 10-year US Treasury actually retreated on Friday.

The call was wrong. In my own account, the strategy over these three weeks was simple: reduce positions and deleverage.

The reasons for the sell-off are well-known

1) Anthropic raised $81.3bn in what is the largest IPO in history, only to see its shares fall 8.5% on its second day of trading. Instead of lifting sentiment around AI, it dragged it down.

2) Doubts about whether AI capex will pay off have spread from big tech to smaller cloud providers, with CoreWeave losing more than half its value in three months. On Thursday, SpaceX issued $32bn in bonds to buy chips, with the largest tranche carrying an interest rate of 8.26%. On the same day, the average rate for investment-grade corporate bonds was 5.95%.

3) After Democrats took control of Congress, Democratic representatives publicly called for a nationwide pause on new AI data center construction last Tuesday.

4) At the Fed, all 13 officials who spoke on rates since Nov 5 leaned toward further hikes. A December rate hike is fully priced in, and Governor Waller is advocating for a 50bps cut.

None of these four points are new; what's new is that they converged in the same week. With the reasons laid out, the real question is just one: where do we go from here? US stock market history is long enough to look up.

Using four rulers to measure thirty years

Extracting the four points above gives us four conditions: a new technology booming; high inflation; continuous Fed rate hikes; credit issues or questioned capital expenditure.

Applying these four rulers to the period from 1991 to 2021, there isn't a single instance where all four aligned. The sticking point is inflation; US inflation hasn't truly been this high in the past three decades. Relaxing this condition, the closest matches are two periods: 2018, and 1999-2000.

Looking at the magnitude of Friday's drop itself: over the past thirty years, the Nasdaq has fallen more than 5.28% in a single day 37 times. Of those, 18 occurred in 2000 and 2001, while 2018 saw zero such drops.

2018: The Fed blinked first

In Q4 2018, the Nasdaq fell 23.6% from its late-August high over 80 trading days. The Fed hiked rates four times that year. In December, no new junk bonds were issued for the entire month, the first time since November 2008.

A month later, the Fed stopped. From the pre-Christmas Eve low, the Nasdaq recovered its highs in four months, gaining 35% for the full year of 2019.

The quick resolution in 2018 was predicated on inflation hovering around 2%, giving the Fed room to maneuver.

2000: The most dangerous part wasn't the April drop

April 14, 2000, was also a black Friday. The March core CPI released that morning rose 0.4% MoM against an expected 0.2%, marking the largest single-month gain in five years. The Nasdaq fell 9.67% that day, wiping out a quarter of its value in a week. Reports at the time mentioned two familiar scenarios: forced liquidations to cover margin calls, and the market having overshot.

Then came the rebound. The Nasdaq hit a low on May 23 and surged 35% by July 17. The S&P 500 was just 0.44% away from its all-time high on September 1. Money rotated from cash-burning internet stocks to profitable leaders like Cisco, Oracle, and Microsoft. The Fed hiked another 50bps in May and then paused.

The turning point arrived in September. Earnings expectations for tech stocks began to be revised downward. On Sept 21, Intel issued a profit warning for Q3 earnings, dropping 21% after hours. The main downtrend started from here. By October 2002, the Nasdaq had lost 77.9% from its peak.

Those who bought the dip in April weren't wrong about the price; they were wrong mistaking a rebound for the end.

The capex cycle of that era is also worth comparing. In 2000, telecom operators' capex reached $106bn, growing 26% annually over four years—more than double revenue growth. Equipment vendors lent customers $25bn-$30bn to buy their own gear. Between 2001 and 2002, approximately $160bn in North American junk bonds defaulted, with over half belonging to the telecom sector.

We stand at another fork in the road, perhaps more like 2000?

Four factors lean toward 2000. Inflation is rising, leaving the Fed with no room to pivot. The reasons for the downturn are intrinsic to AI, not external shocks. Accounting questions have emerged among leaders: back then it was MicroStrategy restating its financials; now it's reports that the SEC is investigating Nvidia's accounting practices. The Nasdaq has seen two single-day drops of over 5% in a month, something that didn't happen at all in 2018.

Two factors lean toward 2018. Valuation: the S&P 500's forward P/E, calculated based on Friday's close, is roughly 16.6x, near its 20-year average, compared to 24.4x in March 2000. Earnings: S&P 500 Q3 earnings grew 31.9% YoY.

But these two are linked. The 16.6x figure assumes earnings estimates aren't revised downward; using trailing twelve-month earnings, the P/E remains at 25.4x. Cisco's earnings in 2000 didn't collapse until 2001.

We stand at another fork in the road, perhaps more like 2000. This doesn't mean the Nasdaq will lose 80%; it means we shouldn't expect the Fed to bail us out within a month like it did in 2018.

Highlighting risks isn't bearishness. Between 1994 and 1999, the Nasdaq experienced nine corrections exceeding 10%, all of which were fully recovered, with the longest taking a year. The tenth correction was in March 2000.

Wall Street is divided

Within 24 hours around Friday's close, there were 17 public statements available: 4 calling it an oversold bounce, 8 saying the selloff is justified or incomplete, and 5 neutral. The cautious side has more voices.

Watch three things going forward

Nvidia. On Dec 1, Jensen Huang testified before three House committees, followed by the company's general counsel the next day. While the hearings were closed-door, what was said—or left unsaid—will leak out. There are no heavyweight earnings reports in the next two weeks unless an exceptionally surprising one emerges, which is low probability. Thus, this hearing is the key driver of AI sentiment for the foreseeable future.

The Fed. How much they hike on Dec 9 and how they guide afterward.

Price action. Can the Nasdaq return to the Oct 6 high of 27,600? In 2018, it returned within four months of bottoming; in 2000, even a 35% rebound failed to get it back.

The first two are calendar events; the third tells us which path we're walking.

[Next Week Outlook]

Thanksgiving week is shortened to three and a half days. Markets closed Thursday, and Friday's early 1 PM close meant light volume, amplifying volatility.

One sentence to recap last week: Nvidia delivered a strong earnings report, and Anthropic completed the largest IPO in history, yet neither could halt the decline. When good news fails to work, the market stops looking at the news.

On this week's agenda: Monday brings the initial Nov PMI and remarks from NY Fed President Williams; Tuesday after-hours features Dell and HP; Wednesday morning sees the Oct PCE and Q3 GDP revision, followed by the Fed's Beige Book in the afternoon. The PCE is the final major inflation data point before the Dec 9 meeting.

Last week tested whether good news could save the market; this week tests whether bad data can still break it.

Buy signals originally relied on two metrics: the equal-weight S&P holding support, and the VIX staying below 30. Both broke last week. Until they recover, we don't discuss buying.

A rebound will come eventually; the key is distinguishing what kind it is.

mou_ding_hou_dong_us_outlook_2026-11-23_29.png

(The above represents personal views and does not constitute investment advice.)

The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.