Snapshot, not a live feed
Every figure below was pulled on and is not updated afterwards. For where the market stands right now, see the Radar.
At the index level Monday was a nothing day. The S&P 500 closed down 0.74% — the 33rd worst of the 251 sessions in the past year, the sort of move that gets no headline. Underneath it, gold fell 3.94%, its eighth-worst single session in five years. Qualcomm fell 7.17%, Boeing 6.91%, Intel 5.67% — while Palo Alto Networks rose 4.63% and Nvidia rose 1.68%. Twenty of the 54 stocks we track went up and 34 went down. The interesting thing is not that the market fell a little. It is how neatly the falling and the rising sorted themselves.
Start here: what one day can and cannot tell you
A single session is the noisiest unit of information in markets. The S&P 500 has closed lower on 25 of the past 251 days by more than 1%, and almost none of those days meant anything a month later. So the useful question about a day like Monday is never “how much did it fall” — it is what sorted the winners from the losers. The size of a day is mostly noise. The ordering of a day sometimes carries information, because it tells you which single risk the market was repricing.
Two terms do the work below. Breadth is simply how many things went up versus down: a market can fall 0.7% because three giant companies fell while everything else held, or because nearly everything fell a bit, and those are different markets wearing the same number. Rate sensitivity is how much a given stock has historically moved on days when the 10-year real yield — the Treasury yield after subtracting expected inflation — moved. We estimate it for each company by regressing its daily returns on the daily change in that yield over the past 248 sessions. A reading of −2.41% for Intel means that over the past year, on an average day when the real yield rose a tenth of a percentage point, Intel fell 2.41%.
The government’s real-yield series has no published observation later than 24 September, when it stood at 2.85% — a five-year high. So this post does not tell you what the real yield did on Monday, because we do not know. What we can do is check whether Monday’s winners and losers were arranged the way a rate move arranges them, using sensitivities measured over a window that ends before Monday. That is a test of pattern, not of cause, and the difference matters.
What actually moved
Three of the ten S&P sectors rose and seven fell, and the whole best-to-worst range was only 1.74 percentage points — a narrow day by sector standards. The three that rose are the three you would name if someone said “defensive”: health care, consumer staples and energy.
Breadth was poor but not extreme. 37.0% of the 54 stocks rose, against a median session of 51.9% over the past year, which puts Monday at the 17th percentile of the year — a bad day for the average stock, though nowhere near the narrowest. That title belongs to 10 October 2025 and 31 December 2025, when just 7% of the list rose.
Two single stocks closed at fresh one-year closing highs — AbbVie and Thermo Fisher — and two at fresh one-year closing lows: Home Depot and McDonald’s. Three of the ETFs also closed at one-year lows: utilities, long Treasuries and high-yield corporate bonds.
The shape underneath: Monday sorted by one number
Here is the day’s six biggest declines with each company’s rate sensitivity printed beside it. The pattern is hard to miss.
Sorting all 54 stocks into thirds by rate sensitivity makes the same point without cherry-picking the extremes. The most rate-sensitive third fell three times as much as the least sensitive third, and only 5 of its 18 members rose.
The formal version: across the 54 stocks, the correlation between Monday’s move and rate sensitivity was +0.413. Because rate-sensitive names carry negative betas, a positive correlation means the sensitive names fell hardest — exactly the arrangement a rising real yield produces. Run that same correlation for every one of the past 251 sessions and the median is −0.04, essentially no relationship. Monday sits at the 85th percentile. On most days, knowing a company’s rate sensitivity tells you nothing about whether it rose. On Monday it told you a good deal.
The same correlation that makes rates the best available explanation for Monday also says rates accounted for 17% of it.
That is the number to hold onto. A correlation of 0.413 squares to 0.17: rate sensitivity explains roughly 17% of the variation in Monday’s moves. The remaining 83% is company news, earnings positioning, index flows and noise — which is why Nvidia rose 1.68% and Applied Materials rose 0.36% despite both sitting in the rate-sensitive half. A pattern can be the clearest thing in the data and still be a minority of it.
The size of the gap was ordinary. The most-sensitive third underperformed the least-sensitive third by 1.25 points. Measured across the year, that ranks at only the 23rd percentile — there have been 57 worse days for rate-sensitive stocks in twelve months, and five of them were worse by more than 3.9 points. Monday’s cross-section was unusually well ordered, not unusually large.
Energy only half confirmed it. Exxon and Chevron carry the rare positive rate betas in this set and both rose, 1.20% and 0.94%. But ConocoPhillips carries the largest positive beta of all, +1.53%, and it fell 0.99%. The energy sector ETF managed +0.10%.
There is a competing explanation. Monday’s moves also correlate −0.25 with each stock’s prior one-month run, meaning recent winners were sold. That is weaker than the rate relationship but not zero, and the two overlap: the rate-sensitive semiconductor names were also the recent winners.
One more caution, and it matters for how you read the word “semiconductors fell”: they did not, uniformly. Qualcomm −7.17%, Intel −5.67%, AMD −3.61% and Micron −2.61% sit alongside Nvidia +1.68%, Applied Materials +0.36%, Texas Instruments +0.09% and Lam Research −0.23%. The sector ETF’s tidy −1.08% hides an 8.9-point spread between its own members.
Gold’s day was the outlier, not the index’s
Gold fell 3.94%, from $393.41 to $377.91 on the GLD share price. Ranked against every session in the five-year window, that is the eighth worst of 1,253 days — the 0.6th percentile. For comparison, the S&P’s 0.74% decline sits at the 17th percentile of its own history. One of those two is a normal Monday.
Gold and long Treasuries falling together, as they did Monday — TLT −0.88%, and at a fresh one-year closing low — is the combination that tends to show up when real yields rise, because both assets compete with a risk-free yield that just got more attractive. It is also the combination that shows up when nothing in particular happens and both simply drift. We are not in a position to tell you which Monday was.
The quarter that ends on Wednesday
Monday is the third-to-last session of the third quarter, so the three-month numbers are nearly final and they are far more dramatic than the day. Energy is up 16.93% since 30 June. Utilities are down 13.43%. That is a 30-point spread inside one index in one quarter.
| Ticker | What it is | 30 Jun | 28 Sep | Quarter |
|---|---|---|---|---|
| XLE | Energy | 53.11 | 62.10 | +16.93% |
| XLV | Health care | 158.66 | 171.26 | +7.94% |
| GLD | Gold | 368.38 | 377.91 | +2.59% |
| SPY | S&P 500 | 746.77 | 765.61 | +2.52% |
| QQQ | Nasdaq 100 | 736.40 | 736.53 | +0.02% |
| RSP | S&P, equal weight | 212.77 | 209.74 | −1.42% |
| IWM | Small caps | 300.45 | 280.02 | −6.80% |
| SMH | Semiconductors | 655.89 | 600.01 | −8.52% |
| TLT | Long Treasuries | 86.42 | 78.62 | −9.03% |
| XLU | Utilities | 45.34 | 39.25 | −13.43% |
The line worth staring at is the fourth against the sixth. The S&P 500 is up 2.52% for the quarter; the same 500 companies held in equal weight are down 1.42%. A 3.94-point gap over three months means the quarter’s gain came from the largest few companies and not from the index’s breadth — and almost all of that gap opened in September.
This is the same split we measured from the bond market’s side in the calm-index, stressed-bond piece, and from the stock side in what already broke in AI. Monday added a data point rather than a new thesis: the index is 1.58% below its one-year closing high of 13 August while 11 of 54 stocks sit in the bottom quarter of their own one-year range.
What this means for your money
A quiet index reading is not a quiet market. Monday’s 0.74% would not register in a monthly statement, yet a Qualcomm holder lost 7.17% and a Palo Alto Networks holder gained 4.63% on the same afternoon. If a portfolio’s day-to-day experience keeps diverging from the headline number, the cause is usually concentration in one of the two ends of this sort.
Several apparently different holdings can be one position. Utilities, real estate, long Treasuries and high-yield debt all fell together and three of them closed at one-year lows. A portfolio holding four of those owns one bet on the direction of interest rates, spread across four tickers — which looks diversified on a statement and behaves as a single position on a day like Monday.
The 17% figure is the part most commentary skips. The rate explanation is genuinely the best one available for Monday’s ordering, and it still leaves 83% unexplained. Anyone offering a single clean reason why the market did what it did on a given day is describing a minority of the variance with total confidence.
Quarter-end numbers get quoted for months. Wednesday’s closes become the figures in every third-quarter report you read in October. On current numbers that report says the S&P gained about 2.5% — and does not say that the equal-weighted version of the same index lost about 1.4%.
None of this is a signal to act on. One session, however legibly arranged, is one session. Our own test of 637 oversold episodes found that even a condition measured over five years bought almost no edge against simply holding the index; a single day’s cross-section buys considerably less. The method behind these figures is on the methodology page.
What would change this read
- The real-yield print for 25 and 28 September. If it shows the real yield flat or lower on Monday, then the rate-shaped cross-section happened without a rate move, and the pattern is coincidence or positioning rather than mechanism. This is the single most informative thing that could arrive.
- Tomorrow and Wednesday. A one-day correlation at the 85th percentile is one draw. If the next two sessions sort the same way, it is a regime; if they scatter, Monday was noise that happened to look tidy.
- Company news we have not accounted for. We measure prices, not announcements. A 7.17% fall in Qualcomm or a 4.63% rise in Palo Alto Networks is the size that usually has a specific cause, and if those causes are company-specific then the rate correlation is partly coincidence riding on two large idiosyncratic moves.
- The beta window. Rate sensitivities are estimated over 248 sessions and most carry low explanatory power individually — for most of these stocks the R² against the real yield is under 0.10. Long Treasuries reach 0.63, which is what a genuinely tight relationship looks like. The thirds test is more robust than any single company’s beta.
- Quarter-end flows. Index funds rebalance into 30 September, which can move prices for reasons with no informational content at all. Some part of Monday and the next two sessions is plumbing.
- −0.74% — the S&P 500 on Monday, the 33rd worst of 251 sessions. An ordinary day.
- −3.94% — gold, its eighth-worst session in five years. The actual outlier of the day.
- +0.413 — the correlation between a stock’s move and its rate sensitivity, against a yearly median of −0.04. The 85th percentile of sessions.
- 17% — how much of Monday’s cross-section that correlation actually explains. The other 83% is everything else.
- 3.94 points — the third-quarter gap between the S&P 500 and the same 500 companies equal-weighted. Nearly all of it opened in September.
Data appendix
All 54 single stocks, Monday’s close-to-close move and the estimated move per +0.10 percentage point day in the 10-year real yield over the 248 sessions ending 24 September 2026. Sorted alphabetically so any claim above can be checked against the whole population rather than a chosen subset.
| Ticker | Monday | Rate beta, per +0.10pt |
|---|---|---|
| AAPL | −0.78% | −0.21% |
| ABBV | +0.73% | −0.35% |
| ADBE | −1.89% | −0.55% |
| AMAT | +0.36% | −1.38% |
| AMD | −3.61% | −1.99% |
| AVGO | −0.92% | −1.33% |
| BA | −6.91% | −1.62% |
| BAC | −2.17% | +0.01% |
| CAT | −0.20% | −0.66% |
| COP | −0.99% | +1.53% |
| COST | +0.02% | +0.09% |
| CRM | −2.88% | −0.15% |
| CVX | +0.94% | +1.09% |
| DE | −0.13% | −0.55% |
| DIS | −0.53% | −0.76% |
| GE | −2.71% | −1.09% |
| GS | −2.05% | −1.06% |
| HD | −1.13% | −1.47% |
| HON | −0.48% | −0.84% |
| INTC | −5.67% | −2.41% |
| JNJ | +0.27% | −0.25% |
| JPM | −1.89% | −0.05% |
| KO | −0.72% | −0.20% |
| LLY | +0.11% | −0.43% |
| LMT | −0.28% | +0.16% |
| LRCX | −0.23% | −1.93% |
| MCD | −1.23% | −0.38% |
| MRK | −0.06% | −0.91% |
| MS | −1.36% | −0.35% |
| MSFT | −1.35% | −0.34% |
| MU | −2.61% | −2.05% |
| NFLX | −2.69% | −0.34% |
| NKE | +1.79% | −1.13% |
| NOW | −3.07% | −0.73% |
| NVDA | +1.68% | −1.24% |
| ORCL | −3.28% | −2.03% |
| PANW | +4.63% | −0.94% |
| PEP | −0.10% | −0.44% |
| PFE | +0.17% | −0.55% |
| PG | +1.91% | −0.50% |
| QCOM | −7.17% | −1.74% |
| SBUX | +0.43% | −0.52% |
| SLB | −0.10% | −0.13% |
| SNOW | −2.33% | −1.00% |
| TGT | +0.62% | −0.46% |
| TMO | +0.53% | −0.82% |
| TSLA | −3.94% | −1.41% |
| TXN | +0.09% | −0.90% |
| UBER | −2.10% | −0.82% |
| UNH | +0.33% | −0.74% |
| UPS | +0.31% | −1.36% |
| V | +0.10% | +0.15% |
| WMT | +0.69% | −0.27% |
| XOM | +1.20% | +1.31% |
Every figure here is a closing price or a statistic derived from closing prices on 28 September 2026, and none of it is refreshed afterwards. Prices from Tradier; yields, inflation expectations, volatility and credit spreads from FRED. Rate betas are estimates from a one-year window and carry low explanatory power for most individual companies. This is a measurement of one trading session, not a forecast, and it contains no recommendation, no price target and no view on what any security will do next. See disclosures.
Newer post
We priced 54,860 box spreads. The guarantee is real. The profit isn't.Every American-style box we priced locked in a loss — all 21,713 of them. The best credible box anywhere beat a Treasury bill by 0.22 percentage points.Earlier post
We tested 637 oversold stocks. The bounce is real. The edge isn't.Oversold stocks rose two-thirds of the time. Six in ten still trailed the index. The signal that paid was never about the stock.Get new posts by email
One email when a new post goes up. No account needed, no other mail, and every message has a one-click unsubscribe.
Educational information only — not investment, tax, or legal advice. Nest Money is not a registered investment adviser or broker-dealer. Market commentary describes what has already happened; it is not a forecast and not a recommendation to buy or sell any security. Options involve substantial risk of loss.