Semis Are Down 31%. After A 30% Nasdaq Drop, 1998 Rose 256% And 2000 Fell 78%.

$NVIDIA(NVDA)$  

Mathematical Money | November 17, 2026


Chip stocks are having their worst stretch of the year, and Nvidia reports tonight. The semiconductor ETF, SMH, closed Monday 31% below its June high. The Nasdaq 100 ETF is 13% below its October high.


Everyone is asking whether this is 2000 again. I lined this selloff up against the two times the Nasdaq fell about 30% in the late 1990s. One ended with the index up 256%. The other ended with it down 78%. The first drop was the same size both times, so the size can't tell you which one this is. Something else can.


Is this the 2000 crash again?


On the chips, it looks a lot like it. The 2000 chip index fell 35% by late May. It then bounced 46%, to within 5% of its high. After that it kept falling for two and a half years and ended 84% below the peak.


SMH has done the first half of that. It fell 25% by late July. It bounced 26% by 5 October, to within 5% of its June high. Now it has turned down and sits 31% below. Count 103 trading days from each peak: the 2000 index was down 28% at that point, and SMH is down 31%.


One more detail rhymes. Intel peaked late in 2000, almost six months after the chip index, and fell 53% in the next 41 days. Nvidia peaked on 6 October, more than three months after the chip ETF, and is down 35% in 41 days.


Does the whole market look like 2000?


Not yet. The Nasdaq in 2000 was 34% below its peak after five weeks. At the same 29 trading days the Nasdaq 100 ETF is 13% below its peak, against 28% for the 2000 Nasdaq. The two indexes aren't identical, so treat that as a rough gap and not an exact one.


So the chips are deep into the story and the rest of the market is still near the start. That is what a selloff in one theme looks like before it spreads. It is also what it looks like before it stops.


Why can't the size of the drop tell me?


In 1998 the Nasdaq fell 29.5% between July and October. People call it a scare now, but it felt like the end of something at the time. The index then rose 256% to its March 2000 peak, and the chip index rose seven times over.


In 2000 the hole was about the same size. The Nasdaq fell 34% in five weeks, then kept going to 78% below its high over the next two and a half years. Two drops of equal depth, with opposite endings.


What was different between them?


The one thing I can find that separates them is what the Fed was doing. In autumn 1998 it cut three times, on 29 September, 15 October and 17 November, down to 4.75%. The Nasdaq's low was 8 October, in the middle of those cuts.


From June 1999 to May 2000 it raised six times, up to 6.5%. The Nasdaq peaked in the middle of that run.


Today futures give about a 96% chance of a hike in December, and at least two more next year, with the 10-year yield at 5.15%. On rates, this looks like 2000 and not 1998. One caution: 1998 had a financial scare behind it and the Fed was responding to it, so a hiking Fed doesn't by itself cause a crash. It's still the cleanest difference I found, and it matches what I found in my Fed post on Friday, where most first hikes left stocks lower three months later.


What is different from 2000?


Valuation, mostly. In March 2000 Cisco traded at around 100 times next year's earnings. Nvidia is nowhere near that on any estimate I found, though the estimates range from about 10 to about 24 times, depending on whose numbers you use. Its reported profit margin is above half of sales, and most dot-com companies at the peak had no profit at all.


What is the same is concentration. The ten biggest stocks are about 36% of the S&P 500 now, against 25% in 2000. The spending behind the AI build is estimated at $660 to $690 billion a year, and nobody can yet say what it earns back.


Where does that leave me?


My view: the chips have already done the first half of what they did in 2000, the wider market hasn't, and the rate path is the swing factor. I'm not calling it a burst. I'm calling it the same position the 2000 chips were in, with the second half still to be decided.


Three paths from here. In the 1998 path the Fed turns toward cuts and chips hold above their 5 November low. In the 2000 path the hikes keep coming and chips keep falling. The marker is 50% down, which the 2000 chip index reached 221 calendar days after its peak. For SMH that is 29 January 2027, and it would need another 28% fall from here. The third path is a slow grind in between.


What would change my mind is a Fed that signals cuts, or chips closing back above the early October bounce high of about $634. Nvidia's numbers tonight and the Fed minutes tomorrow are the next two dated tests.


How do I check any selloff the same way?


You can run this yourself in ten minutes, on any index or stock you own, with free price data. After this you can tell whether a selloff has the 1998 shape or the 2000 shape.


1. Find the peak close and count the trading days since it.

2. Divide every close by the peak close and multiply by 100, so the peak is 100. Lay that over 1998 and over 2000.

3. Read the shape and not the depth. Did it bounce to within about 5% of the high and then fail?

4. Check what the central bank is doing, whether cutting, hiking or holding.


If you have a better way than the Fed to tell 1998 from 2000, I'd like to hear it.


Sources: Tiger daily closes for SMH, QQQ and NVDA through 16 Nov 2026. yfinance daily closes for the Nasdaq Composite, the chip index and Intel, 1998 to 2002. Federal Reserve target-rate dates. Reuters and CME FedWatch for December hike odds, as reported 16 Nov. Cisco, Nvidia, concentration and capex figures are published analyst estimates and differ by source.


$NVDA$ $SMH$ $QQQ$

#semiconductors #Nvidia #stockmarket #investing #FederalReserve

# Nvidia Earnings Meet the Anthropic IPO Window — Where Is the AI Signal?

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Report

Comment(3)

  • Top
  • Latest
  • Heigrace
    ·11-17 21:05
    Should one buy the dip?
    Reply
    Report
    Fold Replies
    • Mathematical Money: 
      sentiments is so bearish that it is hard to predict
      11-17 21:55
      Reply
      Report
  • Jas2davir
    ·11-17 22:09
    So long short is to just hold cash and wait it out?
    Reply
    Report