Two of our five calls broke as the S&P 500 lost 8.6% in a week: Watchlist for Nov. 23
Nvidia fell 22.9% in five sessions and gold 9.2%, while yields dropped. Five research notes for the week of November 23, each with the level that would prove us wrong.
Markets are closed for the weekend, so every price below is as of Friday, November 20 unless it says otherwise. The S&P 500 fund SPY lost 8.6% on the week and closed 11.5% below its peak, on 2.9 times normal volume. At the same time the 10-year Treasury yield fell 0.23 points to 4.92%, from 5.15% a week earlier (U.S. Treasury). Last week we said an invalidation level would tell us when we were wrong. Two of them were crossed. This is the sixth edition of the Weekly Watchlist.
Scorecard: the November 15 ideas
Reference prices are the November 13 closes logged on November 15. Closes are daily closes through November 20 (Yahoo Finance; our arithmetic).
| Idea (Nov. 15) | Nov. 13 close | Nov. 20 close | Change | Invalidation hit? |
|---|---|---|---|---|
| Semiconductors (SMH) | 486.75 | 400.06 | -17.8% | No: needed a close above 518.14 (highest close was 461.90 on Nov. 16) |
| Long Treasuries (TLT) | 79.38 | 81.45 | +2.6% | Yes: needed a close above 81.09; TLT closed at 81.45 on Nov. 20 |
| Gold (GLD) | 355.45 | 322.87 | -9.2% | No: needed a close above 368.78 (highest close was 352.93 on Nov. 16) |
| Consumer discretionary (XLY) | 119.98 | 112.46 | -6.3% | Yes: needed a close under 114.31; XLY closed at 112.46 on Nov. 20 |
| Financials (XLF) | 58.68 | 54.57 | -7.0% | No separate condition; logged with XLY |
| Consumer staples (XLP) | 92.35 | 90.46 | -2.0% | No separate condition; logged with XLY |
| Crude oil (USO) | 145.58 | 139.63 | -4.1% | No: needed a close above 156.06 (highest close was 140.24 on Nov. 17) |
What did not hold: the rotation call. We wrote that cyclical and financial stocks were rising while technology fell, "a rotation out of technology, not an economic retreat," and put XLY's 60-day average, 114.31, as the line that would prove us wrong. XLY closed under it on Friday, and XLF lost 7.0% on the week, nearly as much as the S&P 500. The bond call also failed: we framed it as "yields pushed higher" after the 10-year reached 5.15%. Yields then fell for five straight sessions and TLT crossed above the level we gave. What held, narrowly: the downtrend labels on chips, gold and crude. All three kept falling, and none came near our invalidation levels. SMH's rebound idea was still wrong in practice: the fund did not rebound, it lost 17.8%. Our oil note's "bear" level, a close under 141.45, was crossed on Monday and the fund stayed below it all week.
The backdrop
Growth: the data have not cracked, and the market does not seem to care. Payrolls rose 165,000 in October, with a three-month average of 137,000, and unemployment was 4.1%, down from 4.2% in September (FRED). Initial jobless claims were 203,000 for the week ended November 14, against 202,000 the week before (revised from the 200,000 we cited last Sunday), with a four-week average of 197,750. Nominal retail sales for October rose 0.21% on the month and 6.82% on the year, not adjusted for prices. Industrial production fell 0.43% on the month in October. Housing permits dropped 7.37% to 1.295 million, 8.67% below a year earlier. The dashboard labels growth "accelerating" and inflation "flat"; both are descriptive labels from three-month averages, not forecasts, and they describe October, not this week's market.
Inflation: unchanged since last Sunday, still firm. October headline CPI is 337.476 on FRED, up 0.35% on the month, and core is 339.900, up 0.30%. We still do not quote a 12-month CPI rate: the Bureau of Labor Statistics has no October 2025 reading, so the dashboard's 4.08% compares October with September of last year. September PCE prices rose 0.49% on the month and 3.65% on the year (FRED). Nothing new has been published on CPI this week.
Rates (Treasury par yields, Nov. 20 against Nov. 13). The 10-year fell to 4.92% from 5.15%, the 2-year to 4.63% from 4.88%, and the 30-year to 5.31% from 5.47%. The 10-year minus 2-year spread was 0.29 points, against 0.27 a week earlier. The Fed's own minutes of the October 27-28 meeting, published Wednesday, say that "almost all participants" judged that further action to raise the target range "would likely be appropriate," and that "a couple of participants" said an increase of more than 25 basis points "could warrant consideration at the next meeting" if inflation data were unfavorable (Federal Reserve; our minutes report). The policy range is 3-3/4 to 4 percent. So the bond market moved against the Fed's stated direction. Our reading is that it was bidding for safety; that is an inference, not something the data prove. The FOMC decides on December 9.
Trends (closes through Nov. 20). Up: health care (XLV), consumer staples (XLP), utilities (XLU), long Treasuries (TLT) and the dollar fund (UUP). Down: S&P 500 (SPY), Nasdaq 100 (QQQ), Russell 2000 (IWM), technology (XLK), semiconductors (SMH), energy (XLE), industrials (XLI), gold (GLD), silver (SLV), crude (USO), Bitcoin and Ether. Mixed: financials (XLF), regional banks (KRE), consumer discretionary (XLY), real estate (XLRE) and homebuilders (XHB). "Up" requires the close above its 60-day average and the 20-day above the 60-day; "down" is the reverse. Compared with a week ago, energy moved from "up" to "down" and crude from "mixed" to "down"; financials, regional banks, discretionary, real estate and homebuilders slipped from "up" to "mixed"; long Treasuries moved from "mixed" to "up." SPY is 11.5% below its peak, QQQ 19.9% and XLK 26.2%. Bitcoin was $60,639 and Ether $1,713 on November 22, 37.4% and 49.0% below their peaks, with Bitcoin down 6.6% on the week.
Five ideas to watch
These are research notes on what is changing, not recommendations. Levels are 20- and 60-day averages and a one-day volatility band from the dashboard, not forecasts. Four of the five are funds and one is a single company event read from the company's own filings; we have not done the filings work for other company ideas this week.
1. Nvidia and semiconductors: thesis, the market stopped trading the earnings
What changed: SMH closed at 400.06, down 17.8% on the week, 40.2% below its peak, on 1.99 times normal volume. Nvidia closed at $125.53, down 22.9% from $162.85 a week earlier and 19.3% from the $155.46 it closed at before reporting on Tuesday. Comparison: the company's numbers were strong. Nvidia's results release shows third-quarter revenue of $115.8 billion, up 103% from a year earlier, and a fourth-quarter guide of $130.0 billion, plus or minus 3%, with no China data-center compute assumed. Fourth quarter, though, has an extra week: $130.0 billion over 14 weeks is about $9.29 billion a week, against $8.91 billion in the third quarter's 13, roughly 4.2% more per week (our arithmetic). The same release guides gross margin to 71.0% to 72.0%, against 74.9% in the quarter just reported. Mechanism: a stock can fall after a beat when the debate moves from this quarter's revenue to its quality: margin, the pace per week, and what the company says about its own books. On Friday Reuters reported, citing two unnamed people, that SEC staff are scrutinizing potential accounting issues at Nvidia. Neither the SEC nor Nvidia has confirmed it in anything we have read (our report). Countercase: the selloff was not only Nvidia. Our Friday close report had AMD down 13.2% and Broadcom down 12.2% that day, for reasons we could not verify, and an index-wide move on nearly three times normal volume is hard to pin on one company's filing. Horizon: weeks. Invalidation: SMH closing above its 20-day average (494.49, 23.6% higher). Bull/base/bear: bull, a close above the 419.44 band top; base, 381.57 to 419.44; bear, a close under 381.57, the band floor. Next catalyst: Nvidia's Form 10-Q. Its CFO commentary says the company plans to file "in the coming weeks"; EDGAR shows no 10-Q for the quarter as of Sunday, November 22 (NVIDIA filings). The usual 40-day deadline for a large filer would fall on December 4 (our calculation; we have not confirmed Nvidia's filer category). Its next dividend, $0.25 a share, goes to holders of record on December 2.
2. Long Treasuries: thesis, a flight to safety running against the Fed's stated direction
What changed: TLT closed at 81.45, up 2.6% on the week and in an "up" trend, but still 10.3% below its peak. The 10-year yield dropped from 5.15% to 4.92% and the 2-year from 4.88% to 4.63%. Comparison: the Fed's minutes lean toward more hikes while the bond market moved the other way, and gold, usually the other safe haven, fell 9.2% at the same time. When stocks, bonds' yields and gold all fall together, the common factor is more likely demand for cash and Treasuries than a changed view of inflation (our reading, not a measured result). Mechanism: falling yields lift TLT now. They also lower the benchmark for fixed-rate mortgages and long loans, although Freddie Mac's 7.28% rate on November 19 was set before Friday's rally, so the next weekly reading is the first to show whether it reaches borrowers (our Freddie Mac report). Countercase: September PCE inflation was 3.65% on the year and October CPI rose 0.35% on the month. If the selloff calms, the market could go back to pricing the Fed's hikes and TLT would give back the week. Horizon: days to weeks. Invalidation: TLT closing under its 20-day average (80.99, 0.6% lower). That is close, and we chose it deliberately: this is a fragile call. Bull/base/bear: bull, a close above the 82.03 band top; base, 80.88 to 82.03; bear, a close under 80.88. Next catalysts: weekly jobless claims on Wednesday, November 25 (forecast 200,000 against 203,000), the November jobs report on December 4 (forecast 93,000, unemployment 4.1%, per our markets calendar) and the Fed decision on December 9.
3. Gold: thesis, falling whichever way yields move
What changed: GLD closed at 322.87, down 9.2% on the week, 34.9% below its peak and 2.0% above the low end of its band (316.56), on 2.41 times normal volume. SLV fell 16.2% to 36.97, 65.0% below its peak. Comparison: two weeks ago we tied gold's bounce to falling yields and found it did not fit; last week gold fell as yields rose, and we said that fit; this week yields fell and gold fell harder. Three weeks, three stories, and the explanation we offered failed one of them outright. The dollar fund UUP rose just 0.4% this week, so the dollar does not explain it either (see our gold week report). Mechanism: we have no verified cause. One possibility is that holders are selling gold to raise cash against losses elsewhere, which would fit the volume. That is a hypothesis, not a finding. Countercase: a drop this fast on heavy volume can mark exhaustion as easily as the start of another leg down, and GLD has lost 24.3% in three months, so a sharp bounce would not surprise us. Horizon: weeks. Invalidation: GLD closing above its 20-day average (360.42, 11.6% higher). Bull/base/bear: bull, a close above the 329.30 band top; base, 316.56 to 329.30; bear, a close under 316.56. Next catalysts: the claims report on November 25, the jobs report on December 4 and the Fed on December 9.
4. Cyclicals and financials: thesis, last week's rotation call has become a broad selloff
What changed: XLY lost 6.3% on the week and closed at 112.46, 9.7% below its peak. XLF lost 7.0% to 54.57 and regional banks (KRE) 6.1% to 68.32, 12.3% below their peak. Staples (XLP) lost 2.0% and sit 2.3% below their peak, still in an "up" trend. Comparison: XLP's one-month return is +0.5%, XLY's -2.2% and XLF's -3.7%; the defensive group is holding and the cyclical ones are not. Mechanism: our minutes report describes a staff warning about AI-related debt and banks' indirect exposure to it; we did not re-read that passage for this note, so treat it as attributed to our earlier report. Separately, a bank stock can fall in a stock selloff even when its own loan book has not changed. Countercase: the labor data do not show stress: payrolls +165,000, unemployment 4.1%, claims near 200,000. If that holds, cyclical stocks are being sold with the market rather than for their own reasons. Staples also fell. Horizon: weeks to a quarter. Invalidation: XLY closing above its 20-day average (117.80, 4.7% higher). Bull/base/bear: bull, a close above the 114.70 band top, which would also put it back over its 60-day average of 114.23; base, 110.26 to 114.70; bear, a close under 110.26. Next catalysts: claims on November 25, the jobs report on December 4 and November retail sales on December 16 (all per our markets calendar).
5. Crude oil: thesis, broke below both averages despite war headlines
What changed: USO closed at 139.63, down 4.1% on the week and 14.1% on the month, 17.6% below its peak, now in a "down" trend with volume at 0.92 times normal. January WTI futures closed at $83.41 on Friday (Yahoo Finance). The December contract stopped trading this week, so we do not compare futures prices across the roll; the fund figures above are like for like. Comparison: energy stocks (XLE) fell 6.9% on the week and 10.5% on the month, so equities and crude moved together. The three-month return for USO is still +5.6%. Mechanism: lower crude cuts fuel bills for truckers, airlines and farms and eases headline inflation, which the Fed is weighing. We have not verified why crude fell this week. Our Bahrain report covers the geopolitical risk that has not gone away. Countercase: that risk is the case against a lower price: a supply disruption in the Gulf would hit crude quickly, and the fund's volume was only about normal this week, so it was not a panic move. Horizon: weeks. Invalidation: USO closing above its 20-day average (150.11, 7.5% higher). Bull/base/bear: bull, a close above the 143.92 band top; base, 135.47 to 143.92; bear, a close under 135.47. Next catalyst: Gulf headlines, which are not scheduled; the claims and jobs releases will matter through inflation expectations.
What would change our mind this week
- Nvidia's 10-Q. If it appears and describes an SEC inquiry or an accounting change, our "market stopped trading the earnings" note (idea 1) gets sharper. If it appears with nothing new, the selloff looks more like index-wide fear than a filing problem, and a close in SMH above 419.44 would be the first sign.
- Monday's flash PMI and Wednesday's claims. The composite PMI was 57.6 last time. Claims well above the four-week average of 197,750 would weaken the "data have not cracked" backstop under idea 4 and strengthen the bond call in idea 2.
- The 10-year back above 5.10%. That would mean the flight to safety is over and the Fed's hike path is again in charge, and would invalidate idea 2 and put new pressure on gold.
- The VIX. The volatility index closed at 35.89 on Friday, up from 19.30 a week earlier (Yahoo Finance). A fall back below 25 would mean the market is calming and cyclical stocks (idea 4) have room to recover; a further rise would keep every level above under pressure.
- A holiday week. U.S. markets are closed Thursday, November 26 for Thanksgiving, so Wednesday's session is thin and big moves can come on little volume.
Sources: FRED series CPIAUCSL, CPILFESL, PCEPI, PAYEMS, UNRATE, ICSA, RSAFS, INDPRO, PERMIT (data dates in the text); U.S. Treasury par yield curve through November 20; Federal Reserve minutes of Oct. 27-28; NVIDIA filings on SEC EDGAR, including the Nov. 17 Form 8-K and Exhibit 99.1; Reuters report on Nvidia (as covered in our linked story, unconfirmed); Yahoo Finance daily closes; Chronicle research dashboard built November 22, 2026 at 3:00 PM ET (ETF closes through November 20; Bitcoin and Ether as of November 22). Percent changes and levels are our arithmetic. Not investment advice.
The Weekly Watchlist is research and commentary for a general audience, not personalised investment advice. Do your own research. The Chronicle discloses any position its writers hold in a company named here; none were disclosed for this edition.
Want your business to be the answer?
Get a full package of articles about your business, built so customers, Google and AI assistants can find you.
New ebook from The Company ChronicleHow to Research Like a Wall Street AnalystFind original sources, analyze trends, and build your own insights. 36 pages with worksheets.Get the ebook · $29Instant PDF download

