Vol. III · No. 196 | Thursday, September 24, 2026

Golden Terminal

Overnight into Thursday the selling sits in the same places it sat on Wednesday. The S&P premarket proxy is offered about 0.52%, the Nasdaq proxy 0.92%, the Dow proxy 0.29% and the small-cap proxy 0.27%. Technology is the largest sector move at 1.04% offered and the chips carry it: Advanced Micro Devices 2.08%, Micron 1.95%, Broadcom 1.26%, Nvidia 1.15%. Meta Platforms is offered 1.88%, the largest premarket decline on the roster, the morning after its product event. Energy is bid 0.64% with the refiners leading, Marathon Petroleum 0.93% and Phillips 66 0.64%. The long-bond fund is offered 0.55%, so the yield move has not finished. No sector fund is drifting against its own light by more than sixty basis points and no contradiction flag fires. The market-risk gauge is yellow this morning and the S&P proxy is offered against it, which gets named because that gauge always gets named. Consumer Discretionary and the dollar proxy recorded no premarket trade at all, which is recorded as no trade and not as zero. Six sector funds printed on a few hundred to a few thousand shares and are indicative only. Drift never moves a completed-bar verdict, and none of this does.

The ten-year Treasury yield closed at 5.11%, the highest since July 2007. It rose fifteen basis points in one session, its biggest single-day move in about eighteen months. The five-year and the seven-year each rose sixteen. The thirty-year rose eleven, to 5.40%. The ten-year had closed within five basis points of 4.96% on every one of the five previous sessions. The shape of the curve barely changed: two-year to ten-year went from twenty-five basis points to twenty-six. What moved was the level of the whole thing.

This letter was right about Technology on Tuesday and wrong about Health Care, and the second one matters more. Tuesday's issue called Technology's record 204-point cushion a mechanical problem and said the arithmetic would eventually turn that green light yellow. It went yellow the next bar. The route was wrong: the average did not climb into the reading, the reading fell 43.8 points into the average. In the same issue this letter named Health Care's 7.4-point cushion as the green most likely to fail first. Health Care held and widened to 17.4. The greens that broke were the two widest on the board. The rule that thin cushions go first got the ranking backwards on the one session that tested it.

Five sector lights went down a step and one went up. Technology and Consumer Discretionary fell from green to yellow. Utilities and Real Estate fell from yellow to red. Communication Services fell two steps, green to red, on a 99.2-point decline. Energy was the only upgrade, red to yellow. The market-risk gauge held green for two bars, its longest run of one color this month, and lost it on the third. The board reads two green, five yellow and four red, against five, four and two on Tuesday.

The cause was a growth number nobody had positioned for. The September flash purchasing managers' surveys printed 58.4 composite against a 55.2 estimate, 58.7 services against 56.0 and 57.0 manufacturing against 53.6. All three were forecast to come in below August. All three came in above it. Yesterday's issue asked whether the data would confirm the deceleration or refuse it. It refused it, and the five-year note auction that afternoon stopped at 5.033% against a 4.393% stop at the last sale of the same security, a jump of sixty-four basis points between auctions.

Energy was the only sector that closed higher, and the barrel rose on a build. Crude gained better than three percent and ended a seven-session losing streak in a week the Energy Information Administration reported a 2.97 million barrel crude build against an expected draw. Refinery utilization fell another 2.8%, crude runs fell, and gasoline drew down 1.69 million barrels against an expected build. The Energy Secretary said this morning the administration is weighing diesel export restrictions instead of the outright ban the President backed on Tuesday, which is one feed so far and is carried here as one.

XLV▲ XLI▲ XLK▶ XLE▶ XLB▶ XLP▶ XLY▶ XLC▼ XLF▼ XLRE▼ XLU▼

A growth report nobody expected sent the bond market to its worst day in about a year and a half. Five lights on this board changed color for the worse, and energy was the only sector that closed green.

The index fell seven tenths of a percent and that was the least interesting thing that happened on Wednesday. A September flash growth report that every forecaster had marked down came in higher on all three lines, the bond market repriced the entire Treasury curve upward by six to sixteen basis points in a single afternoon, and five of the eleven sector lights on this board dropped a step while the market-risk gauge fell out of green. The S&P 500 proxy closed at $767.81. The Nasdaq proxy lost 0.84% one day after a second straight record. The small-cap proxy lost 1.84%, the worst of the four, which is what happens to the companies that borrow at the short end when the short end moves fourteen basis points.

Twenty-one of the sixty-seven companies on this roster finished higher and forty-six finished lower. The roster averaged a loss of 0.723% against an index that lost 0.72%, a match to two decimal places, one session after the roster beat the index for the first time in four days. Three sectors went zero for six: Health Care, Real Estate and Utilities. Technology went one for seven. There was no place to hide inside the roster on Wednesday except the oil patch, where four of six names finished higher and the sector average was the best on the board.

The damage concentrated where the arithmetic said it would. The two worst sector funds were the two bond proxies. Utilities lost 1.92% and Real Estate lost 1.55%, and both were downgraded to red. The long-bond fund lost 1.58%. A regulated utility and a landlord are each, stripped down, a contracted stream of payments discounted at the long rate, and on Wednesday the long rate moved further in one afternoon than it had in the previous week. Nothing about either business changed between Tuesday and Wednesday. The discount rate changed.

There is a serious case that Wednesday was a rate shock and not a change of regime, and the desk read here for the price side put it in print the same morning. Stated the way his own readers would recognize it: yes, only 26% of S&P members sit above their twenty-day average and only 30% above their fifty-day, and those are poor numbers. But the Nasdaq Composite closed at an all-time high on Tuesday, the S&P 500 sits a fraction of a percent from its own record, seven of the eleven major developed-market indexes are within three percent of all-time highs with the group averaging about three and a third percent off, the seven largest American companies broke out together after roughly a year of going sideways, and Apple printed a record. A twenty-day average is a statement about one month of price and it is supposed to move around. Building a whole market thesis on one short-window indicator is how people talk themselves out of a bull market. That is the case, and it went out before the nine forty-five print and before the one o'clock auction.

This board answers differently, and the reason is that it is not measuring breadth. It measures twenty completed sessions of price in each of eleven sectors plus the index, and on Wednesday seven of those twelve instruments printed a lower reading than the day before. Six of the seven had printed a higher reading on Tuesday. Communication Services lost 99.2 points in a single bar, more than double the largest single-session move in either direction that this board recorded on Tuesday. Technology lost 43.8. Consumer Discretionary lost 41.2. Real Estate lost 34.7. None of that is a participation statistic. It is the same twelve instruments measured the same way on two consecutive days, with one input changed in between.

The number that leaves the screen is the mortgage rate, and it has not caught up yet. The Mortgage Bankers Association put the thirty-year contract rate at 7.12% on Wednesday morning, measured over the week that ended on the eighteenth, before any of this. The ten-year that sets it rose fifteen basis points on Wednesday afternoon. Freddie Mac publishes its own survey at noon today against a 6.95% prior, and it is measuring a week the tape has already left behind. Anybody locking a rate this week is locking against Wednesday's close, not against the number in Wednesday's headline. The survey catches up next Wednesday. The borrower catches up at the closing table.

Brad Hoppmann
Filed from Taintsville, Florida · Pop. < 1,000. 'Taint in the Beltway, 'taint in any backwards corrupt city, just a Florida man with a sharp pencil and a long memory of expensive lessons.

WHAT TO WATCH BEFORE THE OPEN. New York Fed President Williams already spoke overnight and said another rate increase this year is a reasonable expectation, adding that the Fed is finished with explicit forward guidance. Governor Barkin follows at 8:00 and Cleveland's Hammack at 8:50. Initial jobless claims land at 8:30 against a 201,000 estimate and a 196,000 prior, with continuing claims estimated at 1.750 million against 1.730 million and the second-quarter current account at a negative $255 billion against a negative $226.8 billion. New home sales print at 10:00 against a 620,000 estimate, which is the first housing data since the contract rate crossed seven percent. The Kansas City Fed manufacturing index follows at 11:00 with an estimate of 9 against a prior 17. Freddie Mac's thirty-year survey is at 12:00 against a 6.95% prior. The seven-year note auction is at 1:00 against a 4.512% prior stop, and after Wednesday that is the print of the day. Costco reports after the close.

Five lights went down a step. The one this letter said would break first is the one that held.

EARLY EARNINGS UPDATE · DESK NOTE. Three roster names now sit inside the seven-day reporting window, up from two. The nearest reports tonight after the close and its revenue consensus moved higher overnight, the first revenue revision in that series after nine days of tracking and one day after its earnings line moved up a penny. The second reports a week from tonight with both lines unchanged. The third entered the window this morning, reports a week from today, and is the weakest year-to-date name on the roster. Zero roster names report before tonight. Three non-roster names report this morning and eight more inside the following week. No alignment fired on any of the three: the market-risk light changed color, which resets the persistence counter on every name at once. No position language appears anywhere in this issue.

The Reader, Investor & Active Trader Tiers. The complete sector-by-sector breakdown, rotation snapshot, and validation data continue below.

CCI(20) Verdict: YELLOW, as of Wednesday's close · XLK (current +193.5 vs. prior +237.3, 20-day average +44.1) · session -0.47%

YELLOW as of Wednesday's close, downgraded from green after five sessions. The reading fell 43.8 points and the cushion above its own average narrowed from 204.2 to 149.4. The sector fund is offered about 1.04% before the bell, the largest premarket sector move on the board.

Tuesday this letter called the 204.2-point cushion a record and a mechanical problem, and said the average climbing toward the reading would eventually turn the light yellow while the sector was still going up. The light went yellow the next bar and the sector was not still going up. It fell 0.47%, one of seven roster names finished higher, and the reading dropped 43.8 points into an average that barely moved. Being right about the day and wrong about the route should be said out loud on the morning the call lands, because the next person who reads a wide cushion as a warning will want to know which half of that sentence held up. The mechanism that actually fired is duration. A sector whose earnings sit furthest out in time takes the largest present-value hit when the whole curve moves up sixteen basis points in the belly, and the chips led the move down. Technology is still second on the year at plus 34.1% and remains one of only four sector funds above its fifty-day average, which is the trend refusing to confirm the momentum break.

Micron (MU) fell 2.22% to $1,071.88 and lagged its sector fund by 175 basis points. It holds plus 263.2% on the year, still the best figure on the roster by eighty-two percentage points. The company reports Wednesday the thirtieth after the close against a consensus of 31.43 a share on roughly $50.8 billion of revenue, both lines unchanged overnight. Its own momentum reading fell from plus 190.3 to plus 168.9, its first decline of this cycle, which puts it in the deceleration state and not the reversal state. It is offered 1.95% before the bell.

Oracle (ORCL) fell 3.11% to $144.56 and lagged its sector fund by 264 basis points, the third-worst relative move anywhere on the board. It is minus 26.8% on the year, fourth-worst on this roster. No confirmed same-day catalyst turned up across multiple feeds and none is asserted here. Somebody who bought this in January has watched better than a quarter of the position disappear while the index they are measured against gained twelve percent, and Wednesday added three more points to that gap on a day when the story being sold was interest rates rather than software.

Microsoft (MSFT) rose 0.52% to $500.59, the only technology name on the roster to finish higher, and beat its sector fund by 99 basis points. It is plus 3.3% on the year. Its own momentum reading crossed above both its twenty-session and its ten-session average, which closes the reading-ladder ambiguity this letter has been carrying on one name or another for five consecutive runs. There is no open ambiguity on any roster name this morning.

CCI(20) Verdict: YELLOW, as of Wednesday's close · XLE (current -102.1 vs. prior -133.5, 20-day average +48.7) · session +0.96%

YELLOW as of Wednesday's close, upgraded from red and the only upgrade on the board. The reading improved 31.4 points and still sits 150.8 below its own trailing average, which remains the widest dislocation anywhere here. The fund is bid about 0.64% before the bell on real size.

Energy gained 0.96% with four of six names higher and a roster average of plus 1.10%, the best on the board. Crude rose better than three percent and ended a seven-session losing streak, and it did that in a week the Energy Information Administration reported a 2.97 million barrel crude build against an expected 0.6 million draw. The rest of that report explains the apparent contradiction. Refinery utilization fell 2.8% on the week after falling 1.0% the week before, crude runs fell by more than half a million barrels a day, and gasoline drew down 1.69 million barrels against an expected build. Refiners are processing less, so crude accumulates and products do not. The Energy Secretary said this morning that the administration is looking at diesel export restrictions instead of the ban the President endorsed on Tuesday, which narrows the policy but does not remove it, and that narrowing currently rests on one feed. Energy is first on the year at plus 39.5% and one of four funds above its fifty-day average.

ConocoPhillips (COP) rose 2.25% to $128.09, the best energy name on the roster, and beat its sector fund by 129 basis points. It is plus 36.8% on the year. A pure producer with no refining to speak of is the cleanest read on the crude price in this group, and the crude price stopped falling on Wednesday for the first time in eight sessions. It is bid 0.77% before the bell.

Exxon Mobil (XOM) rose 1.59% to $161.23 and holds plus 34.3% on the year. It beat its sector fund by 63 basis points. An integrated major owns production, refining and chemicals on one balance sheet, which makes it the least exposed name in this group to the specific question of what Washington does to the price of diesel, and on Wednesday it went up with the barrel while the pure refiners did not.

Marathon Petroleum (MPC) fell 0.33% to $388.38, and together with Phillips 66 at minus 0.12% the two refiners were the only energy names on this roster to finish lower on a day crude rose better than three percent. It is plus 138.5% on the year, third-best here. A refiner earns the spread between crude and product, so a rising barrel is a cost increase unless the product price rises with it, and an export restriction caps the product price by policy. The weekly utilization number falling 2.8% is that squeeze showing up in operating decisions. It is bid 0.93% before the bell, the largest premarket gain on the roster.

CCI(20) Verdict: RED, as of Wednesday's close · XLF (current -175.5 vs. prior -173.1, 20-day average -79.2) · session -0.47%

RED as of Wednesday's close, held for a second session. The reading fell 2.4 points, the smallest move on the board, after Tuesday's 42.6-point collapse. The fund is bid about 0.05% before the bell.

On Tuesday a news wire attributed bank weakness to a flattening yield curve and the closing curve steepened. On Wednesday the curve steepened by another basis point while the level of the entire thing rose, fourteen at the two-year and fifteen at the ten. A deposit-funded lender pays a rate that reprices slowly and earns a rate that reprices with new assets, so a parallel lift of the whole curve is a widening of what it makes, spread over time. The group fell 0.47% anyway, with two of six names higher, and the momentum reading moved 2.4 points, which is this board saying nothing new after saying something violent the day before. Financials is eighth on the year at minus 0.5% and below its fifty-day average.

Berkshire Hathaway (BRK.B) rose 0.73% to $507.17, the best financial on the roster, and beat its sector fund by 120 basis points. It is plus 1.2% on the year. An enormous cash position is a drag when money is free and a business when it is not, and the two-year note now pays 4.85%, up fourteen basis points in a session. Yesterday the argument for owning it depended on what happens to short rates next. On Wednesday short rates answered.

Mastercard (MA) rose 0.72% to $559.92 and beat its sector fund by 119 basis points, the second-best relative move in the group. It is minus 1.9% on the year. A payments network takes a toll on nominal spending, and a 58.4 composite purchasing managers' reading against a 55.2 estimate is a description of nominal spending accelerating. It was among the few names anywhere on this roster that Wednesday's growth print helped rather than hurt.

Goldman Sachs (GS) fell 1.38% to $936.36, the worst financial on the roster, and lagged its sector fund by 91 basis points. It is plus 5.9% on the year. An investment bank earns underwriting and advisory fees on issuance, and issuance gets harder the week a five-year Treasury auction stops sixty-four basis points above the last one. When the risk-free borrower has to pay up, everybody behind it in the queue pays up more.

CCI(20) Verdict: RED, as of Wednesday's close · XLC (current +13.8 vs. prior +113.0, 20-day average +50.1) · session -0.85%

RED as of Wednesday's close, downgraded two steps from green in a single session. The reading fell 99.2 points, more than double the largest single-bar move this board recorded on Tuesday in either direction. The fund is offered about 0.36% before the bell on roughly 5,800 shares and is indicative only.

Tuesday this letter wrote that Monday's eleven percent move in a single dominant constituent had lifted the series far enough that one broadly negative session could not pull the current reading under its own average. Wednesday tested that and the answer was no. Three of six names finished higher, which is better breadth than Tuesday's zero of six, and the reading still fell ninety-nine points, because a twenty-session momentum measure keeps giving back what a single outsized bar lent it. Alphabet falling 3.80% with no confirmed catalyst is most of the rest of the arithmetic. A two-step downgrade inside one session is the sharpest verdict change this board has produced this month, and it happened on a day the sector fund fell less than one percent. Communication Services is ninth on the year at minus 4.7% and remains one of four funds above its fifty-day average.

Alphabet (GOOGL) fell 3.80% to $337.83, the worst name in the group, and lagged its sector fund by 295 basis points, the second-worst relative performance on the board. It is plus 6.6% on the year. No confirmed same-day catalyst was located across multiple independent feeds and none is offered here. A single position of this size moving nearly four percent is most of what happened to the sector reading, which is a reminder of how much of this group's verdict depends on two companies.

Meta Platforms (META) rose 1.02% to $744.10, the best name in the group, and beat its sector fund by 187 basis points, the strongest relative move anywhere on this board. It is plus 12.3% on the year. The company held its Connect conference on Wednesday evening and introduced an expansion of its Muse assistant, a palm-sized handheld device that carries it, virtual-reality glasses priced at $1,299 and a camera-free audio version of its Ray-Ban line. It is offered 1.88% before the bell, the largest premarket decline on the roster, which one outlet attributes to profit-taking after this week's run. That attribution is a single feed and is carried as one.

Netflix (NFLX) fell 1.11% to $71.36 and remains minus 24.2% on the year, fifth-worst on this roster. It beat its sector fund by 26 basis points, which after nine months of this is the smallest kind of good news and is still the first time in weeks it has led its own group on a down day. What would change the read is that happening more than once. One session is not a pattern, and on Wednesday it was the only thing in the column.

CCI(20) Verdict: YELLOW, as of Wednesday's close · XLY (current -88.2 vs. prior -47.0, 20-day average -123.4) · session -1.50%

YELLOW as of Wednesday's close, downgraded from green after two sessions. The reading fell 41.2 points and the cushion above its own average narrowed from 70.6 to 35.2. The fund recorded no premarket trade before the bell, which is recorded as no trade and not as zero.

Last Thursday this sector turned green on a cushion of three tenths of a point and reversed in a single bar. Monday it turned green again on 43.4 points and Tuesday widened that to 70.6, and this letter said at the time that the size of the cushion was the distinction to watch. The larger cushion bought two sessions instead of one. That is the whole of what the distinction was worth on this test, and it is smaller than the confidence the sentence carried. One of six names finished higher. The sector is last of eleven on the year at minus 7.9% and trades below its fifty-day average, so the momentum improvement that produced the green in the first place was always running against the trend.

McDonald's (MCD) fell 4.81% to $238.32, the worst name on the entire sixty-seven-name roster, and lagged its sector fund by 331 basis points, the worst relative performance on the board. It is minus 21.9% on the year. The company held an analyst and investor day on Wednesday and laid out a ten-year plan including roughly $8.5 billion of restaurant investment, confirmed on the posted transcript and independent coverage. Somebody who has held this since January watched management put a decade of strategy on the table and watched the market take another 4.81% off in the same afternoon. That combination is a harder thing to sit with than a quiet decline, because it removes the excuse that nobody explained the plan.

Nike (NKE) fell 0.14% to $36.05 and remains minus 43.7% on the year, the worst performer on this roster by six percentage points. It entered the seven-day reporting window this morning and reports on Thursday the first against a consensus of 0.4383 a share on roughly $11.3 billion of revenue. The thing anyone holding it should have in front of them is this: the company has beaten the consensus earnings estimate in each of its last three reported quarters, including 0.72 against a 0.11 estimate in June, and the stock is down better than forty-three percent anyway. Being right about the operating business and wrong about the stock for nine months is its own particular expense. Analysts carry a consensus target of $47.62 with a high of $75 and a low of $30, and that low estimate sits below Wednesday's price.

Home Depot (HD) fell 2.83% to $296.72 and lagged its sector fund by 133 basis points. It is minus 13.6% on the year. A home-improvement retailer is a bet on housing turnover and on owners borrowing against the equity in a house they already have, and both of those get harder when the ten-year prints a nineteen-year high. Tuesday the market bought this stock 2.74% higher on the morning the contract rate crossed seven percent. Wednesday it took all of that back and a little more.

CCI(20) Verdict: GREEN, as of Wednesday's close · XLI (current -59.7 vs. prior -69.9, 20-day average -109.9) · session -0.10%

GREEN as of Wednesday's close, held for a third session. The reading improved 10.2 points and the cushion above its own average widened to 50.2. The fund is offered about 0.15% before the bell on roughly 1,900 shares and is indicative only.

Industrials posted the second-best roster average of the day at plus 0.54% with five of six names up, on a session the fund itself lost a tenth of a percent and the index lost seven tenths. It is now one of only two green verdicts on this board. The flash manufacturing survey printing 57.0 against a 53.6 estimate is the cleanest piece of evidence sitting under that verdict, and it is the same print that wrecked the bond market, which is the honest shape of Wednesday: the news was good for the things that make things and bad for the things that borrow. The tension here is that the sector trades below its fifty-day average while carrying a green momentum reading, so this is momentum improving underneath a broken trend. Industrials sits fourth on the year at plus 9.3%, having passed Health Care on Wednesday.

Boeing (BA) rose 1.12% to $199.93, the best industrial on the roster, and beat its sector fund by 122 basis points. It is minus 8.4% on the year. Yesterday it was the worst name in this group. Aircraft are sold on multi-year order books to airlines financing in the corporate bond market, so a fifteen basis point move in the ten-year is a slow poison rather than a same-day event, and Wednesday's move was about orders and not about rates.

Deere (DE) rose 0.90% to $709.48 and beat its sector fund by 100 basis points. It is plus 52.2% on the year, fifth-best on this roster. Farm equipment is bought with borrowed money against expected crop receipts, and the five-year note, which is roughly where that paper prices, rose sixteen basis points on Wednesday. The stock went up anyway on the same morning, which says the order book is doing more work than the financing cost. It is offered 0.49% before the bell.

Caterpillar (CAT) rose 0.50% to $812.02 and holds plus 40.6% on the year. It entered the roster's top seven year-to-date performers on Wednesday, displacing Iron Mountain, which is the first change in that list this week. Heavy equipment is the most direct expression on this roster of a manufacturing survey that printed three and a half points above estimate. It is offered 0.96% before the bell.

CCI(20) Verdict: YELLOW, as of Wednesday's close · XLB (current -78.2 vs. prior -81.7, 20-day average -75.8) · session -0.49%

YELLOW as of Wednesday's close, held. The reading improved 3.5 points and now sits 2.4 below its own average, the narrowest gap between any reading and its average anywhere on this board. The fund shows about minus 0.74% before the bell on roughly 140 shares and is indicative only.

Materials is closer to a green verdict than any other yellow or red light here, and it got there by improving 3.5 points on a session the fund fell half a percent. Two of six names finished higher. Yesterday this letter flagged the metals complex being sold before the bell as a caution that sat below the contradiction threshold and named it anyway. The caution was right. Silver lost 4.23% on the session, gold lost 1.80%, Newmont lost 2.92% and Freeport lost 2.38%, and silver is offered another 1.41% this morning. The mechanism is the same one that hit the utilities: a dollar proxy that rose 0.60% and a nominal yield that rose fifteen basis points together raise the cost of holding a metal that pays nothing. Materials is third on the year at plus 10.5% and below its fifty-day average.

Linde (LIN) rose 1.03% to $469.72, the best material on the roster, and beat its sector fund by 152 basis points. It is plus 10.2% on the year. Industrial gases are sold on long contracts to refiners, chemical plants and chip fabricators, which makes this the one name in the group whose revenue does not move with a metal price, and on a day the metals were sold that was the whole difference.

Newmont (NEM) fell 2.92% to $123.55, the worst material on the roster, and lagged its sector fund by 243 basis points. It is plus 22.3% on the year. The gold proxy lost 1.80% and the miner lost nearly three, which is the leverage in this business working the direction owners do not want. Tuesday it was the best name in this group on the same leverage running the other way.

Freeport-McMoRan (FCX) fell 2.38% to $72.58 and holds plus 40.3% on the year, seventh-best on this roster. Copper is the industrial metal tied most directly to electrical infrastructure, so a manufacturing survey three and a half points above estimate argues for it while a fifteen basis point jump in the discount rate argues against it. On Wednesday the discount rate won. It is offered 1.05% before the bell.

CCI(20) Verdict: GREEN, as of Wednesday's close · XLV (current -7.0 vs. prior -9.8, 20-day average -24.4) · session -0.64%

GREEN as of Wednesday's close, held for a second session. The reading improved 2.8 points and the cushion above its own average widened from 7.4 to 17.4. The fund shows about minus 0.07% before the bell on roughly 1,900 shares and is indicative only.

Yesterday this letter named the 7.4-point cushion under this verdict as the thinnest green on the board and the one most likely to fail first. It widened to 17.4. The greens that failed were Technology's, the widest this engine has produced all year, and Communication Services, which fell two steps in one bar. The stated rule was that a thin cushion is fragile. What Wednesday showed is that a wide cushion built on a single fast-moving constituent is more fragile than a thin one built on a sector that grinds, and that is the version this desk carries forward. Zero of six roster names finished higher here and the light still improved, which is the same twenty-session arithmetic Communication Services demonstrated on Tuesday running the other way. Health Care is fifth on the year at plus 9.0% and one of four funds above its fifty-day average.

UnitedHealth (UNH) fell 0.45% to $371.29, the best relative performer in the group, and beat its sector fund by 19 basis points. It is plus 12.2% on the year. A managed-care company is a spread business on medical costs and the least rate-sensitive structure in this sector, which on a day defined by the discount rate is the reason it held up.

Eli Lilly (LLY) fell 1.64% to $1,150.99, the worst name in the group, and lagged its sector fund by 100 basis points. It is plus 6.9% on the year. A pharmaceutical company whose valuation rests on a long pipeline of future cash flows carries the same duration problem the chips carry, and it got the same treatment on Wednesday. No confirmed same-day company catalyst was located.

Intuitive Surgical (ISRG) fell 0.94% to $398.30 and remains minus 29.7% on the year, third-worst on this roster. Tuesday it rose a tenth of a percent and this letter said that was not a recovery. Wednesday it gave that back and more, in a sector whose light held green. A stock that cannot participate when its own group is the second-strongest thing on the board is telling you something the sector verdict does not. Surgical robotics is capital equipment sold into hospital budgets, and hospital budgets are financed too.

CCI(20) Verdict: YELLOW, as of Wednesday's close · XLP (current -99.1 vs. prior -102.0, 20-day average -86.2) · session -0.36%

YELLOW as of Wednesday's close, held. The reading improved 2.9 points and remains 12.9 short of a green verdict. The fund is bid about 0.16% before the bell on roughly 1,700 shares and is indicative only.

On Tuesday this sector produced the largest single-bar improvement on the board at 47.9 points and went six for six. On Wednesday it added 2.9 more with three of six names higher. A second consecutive improvement off a deep base counts for something, and it arrived at about one sixteenth the size of the first, which is what a one-day rotation into the laggards looks like when the rotation stops. Groceries, household products and tobacco are the least rate-sensitive revenue streams on this roster, and on a day the discount rate did all the work this group finished third-best of eleven by doing almost nothing. Staples is sixth on the year at plus 6.1% and below its fifty-day average.

Costco (COST) rose 0.59% to $904.70, the best staple on the roster, and beat its sector fund by 95 basis points. It is plus 5.1% on the year. The company reports after tonight's close against a consensus of 6.54 a share on roughly $95.0 billion of revenue. The revenue estimate moved up by about $103 million overnight, the first revenue revision in this series across nine days of tracking, arriving one day after the earnings line moved up a penny. Analysts carry a consensus target of $1,101.78 against Wednesday's close, a gap of 21.8%. Its own momentum reading now sits above both its twenty-session and ten-session averages.

Coca-Cola (KO) fell 0.59% to $88.09 and holds plus 26.1% on the year, the best figure in this sector and among the better ones on the whole roster. It is bid 0.45% before the bell. A global beverage business with pricing power and a dividend is the most defensive structure in a defensive sector, and it has outperformed the index by fourteen points this year while almost nobody discussed it.

PepsiCo (PEP) fell 0.77% to $130.18 and is minus 9.1% on the year, the weakest name in this sector and tenth-worst on the roster. A snack and beverage company sells into the household most squeezed by a diesel-driven rise in the cost of moving goods, and it carries the freight cost on the way in as well. That is a margin problem on both ends and the stock has spent nine months reflecting it.

CCI(20) Verdict: RED, as of Wednesday's close · XLU (current -176.3 vs. prior -154.6, 20-day average -120.5) · session -1.92%

RED as of Wednesday's close, downgraded from yellow after one session. The reading fell 21.7 points and is again the deepest number on this board. The fund is bid about 0.18% before the bell on roughly 23,000 shares.

Utilities lost 1.92%, the worst of the eleven, zero of six roster names finished higher, and the sector roster average of minus 1.99% was the weakest on the board. Yesterday it climbed out of red after nine sessions. Today it is back, one bar later, which is the reversal pattern named here before as the most common way this engine produces a false upgrade, and it is the second time in eight sessions that pattern has fired. The mechanism is arithmetic. A regulated utility earns an allowed return on a rate base it finances with long-dated debt, and the long end rose eleven to sixteen basis points across the curve in a single afternoon after sitting still for three sessions. Utilities is tenth on the year at minus 7.3% and below its fifty-day average.

NextEra Energy (NEE) fell 2.83% to $77.02, the worst utility on the roster, and lagged its sector fund by 91 basis points. It is minus 4.3% on the year. A utility with a large renewable development arm funds construction years ahead of the revenue it produces, which makes it the most rate-sensitive structure in a rate-sensitive sector. It is bid 0.25% before the bell.

NRG Energy (NRG) fell 1.98% to $100.72 and is minus 37.7% on the year, second-worst on this roster. A merchant generator sells power into wholesale markets with no regulated return, so it should be the one utility that benefits from an economy printing a 58.4 composite survey. It fell with the group anyway. Anyone who bought this for the data-center electricity thesis has now lost better than a third of the position while the thesis itself kept getting louder, and Wednesday did nothing to close that gap.

American Electric Power (AEP) fell 1.55% to $118.40, the best relative performer in the group, and beat its sector fund by 37 basis points. It is plus 2.6% on the year. A pure regulated utility with no merchant exposure is the most boring structure in this sector, and on a day every name in the group fell, boring lost the least. It is bid 0.35% before the bell.

CCI(20) Verdict: RED, as of Wednesday's close · XLRE (current -145.7 vs. prior -111.0, 20-day average -115.0) · session -1.55%

RED as of Wednesday's close, downgraded from yellow. The reading fell 34.7 points, moving from 7.5 short of a green verdict to 30.7 below its own average in one session. The fund is bid about 0.16% before the bell on roughly 650 shares and is indicative only.

On Tuesday this sector sat 7.5 points from a green verdict, the narrowest gap of any non-green light on the board. On Wednesday it lost 34.7 points and went red, a forty-two point round trip in one session, and zero of six names finished higher. Tuesday's issue argued that the commercial landlords on this roster are financed in the corporate bond market at the long end and that the long end had not moved. The long end moved. The argument was correct about the mechanism, and a correct mechanism cuts both ways the moment its input changes, which is the difference between a mechanism and a forecast. Real Estate is seventh on the year at plus 3.7% and below its fifty-day average.

Simon Property Group (SPG) fell 0.87% to $204.16, the best relative performer in the group, and beat its sector fund by 68 basis points. It is plus 10.7% on the year. A mall owner collects rent plus a percentage of tenant sales, which gives it a slice of the same nominal spending the flash surveys just described accelerating, and that slice is the only thing in this sector that gets better when the economy runs hot.

American Tower (AMT) fell 2.55% to $170.78 and lagged its sector fund by 100 basis points. It is minus 2.6% on the year. A tower company leases space on structures it built decades ago under contracts that escalate at a fixed rate, which makes it the purest bond substitute on this roster and the name most mechanically exposed to Wednesday. It recorded no premarket trade before the bell.

Iron Mountain (IRM) fell 2.57% to $115.09, the worst name in the group, and lagged its sector fund by 102 basis points. It is plus 38.7% on the year and it dropped out of the roster's top seven year-to-date performers on Wednesday as Caterpillar passed it. A document-storage business that converted itself into a data-center operator carries both the duration of a landlord and the enthusiasm of a technology name, and on Wednesday both halves were sold. It recorded no premarket trade.

S&P 500 proxy SPY at $767.81, plus 12.0% on the year, for reference. Market-risk gauge YELLOW at plus 57.9 against a minus 39.6 average, downgraded from green after holding it two bars.

Roster breadth was 21 higher and 46 lower out of 67, with none unchanged. The roster averaged minus 0.723% against an index that lost 0.72%, matching it to two decimal places one session after beating it for the first time in four days. Three sectors went zero for six: Health Care, Real Estate and Utilities. Technology went one for seven. Energy at four of six and Industrials at five of six were the only groups with real participation to the upside. Four of eleven sector funds still trade above their fifty-day averages, the same four as Tuesday: Technology, Communication Services, Energy and Health Care. That count did not change on a session with five verdict downgrades, which means momentum broke on Wednesday and trend has not, yet.

Best relative performers against their own sector funds: Meta Platforms at plus 187 basis points, Tesla at plus 182, AT&T at plus 165, Linde at plus 152, Nike at plus 136, ConocoPhillips at plus 129, Boeing at plus 122, Berkshire Hathaway at plus 120 and Mastercard at plus 119. Worst: McDonald's at minus 331, Alphabet at minus 295, Oracle at minus 264, Newmont at minus 243, Broadcom at minus 215, Freeport-McMoRan at minus 189, Micron at minus 175 and Home Depot at minus 133. The spread between the best and worst year-to-date performers on this roster narrowed to 306.9 percentage points from Tuesday's 315.0, and it narrowed because the leader fell rather than because the laggard recovered.

The financial channels this morning have a Trump-Xi summit, a two-month trade truce and a new pocket-sized artificial-intelligence gadget to talk about, and they will spend the day on those three. The actual news was a purchasing managers' survey nobody was watching, a five-year auction that stopped sixty-four basis points wide, and eleven momentum lights that answered both within the hour.

Three roster names in the window, up from two. Nike entered this morning and is the roster's weakest year-to-date name. Costco's revenue consensus rose about $103 million overnight, the first revenue revision in nine days of tracking; Micron carried both lines unchanged. Estimates are sell-side consensus figures and are labeled as such. Report times marked unconfirmed were not verified on a second independent feed.

Caterpillar entered this list on Wednesday and Iron Mountain left it. The spread between the best and worst name narrowed to 306.9 percentage points from 315.0, and it narrowed because Micron fell 2.22% while Nike fell 0.14%, which is a leader coming down and not a laggard coming back. The right-hand column is the one that needs the sentence this morning. Nike reports a week from today. It has beaten the consensus earnings estimate in each of its last three reported quarters and it is still the worst name on this roster by six percentage points, which is what a market looks like when it has stopped grading a company on its quarter and started grading it on something else.

Wednesday had one cause and it is not in dispute. The September flash growth surveys came in far above estimate on all three lines when all three were forecast to slow. The bond market read that as a Federal Reserve with more work in front of it, the five-year auction that afternoon stopped sixty-four basis points above the last sale of the same security, and every Treasury maturity from one year out closed between six and sixteen basis points higher. The ten-year finished at 5.11%, its highest since July 2007. Everything on this board that gets valued by discounting a future stream of money fell, and the two sectors built almost entirely out of that structure, the utilities and the landlords, fell hardest and were both downgraded to red.

The board lost five lights and gained one. Energy was the upgrade and the only sector that closed higher. Industrials and Health Care are the two greens left standing, and they are the two whose readings improved on a day the index fell. The market-risk gauge held green for two consecutive bars, its longest run of a single color this month, and gave it up on the third. Nine sessions ago this board was four green, three yellow and four red. Two sessions ago it was five, four and two. This morning it is two, five and four. A board that moves that far in forty-eight hours is measuring something real and measuring it on a short clock, and both halves of that matter.

The case against this read was made in print the same morning and it has not been answered. A twenty-day participation reading describes one month of price. The Nasdaq Composite closed at a record on Tuesday, most of the world's large developed markets sit within a few percent of all-time highs, and the biggest American companies broke out together after a year of nothing. Wednesday was one session and one rate print. Note also the shape of the room: of the four desks whose work sits behind this page, three now read the tape bearishly, and a crowded bear is a worse bear. This board joined that side on Wednesday. It joined on arithmetic and not on opinion, and that is the most this letter will claim for it.

The bill for Wednesday gets written in two places. The first is the Treasury's. Sixty-four basis points on five-year paper is $640 million a year for five years on every $100 billion issued at that stop, and that obligation locked the moment the auction closed, payable by people who were not watching. The second is the closing table. The Mortgage Bankers Association put the thirty-year contract rate at 7.12% on Wednesday morning, measured through the eighteenth, before any of this happened. Freddie Mac prints its own survey at noon today against a 6.95% prior, measuring a week the tape has already left behind. Nobody's income moved on Wednesday afternoon. The rate that prices a house moved fifteen basis points.

Two prior tests need settling. Issue 194's two-week clock is still running with eight trading days left: a new all-time high on the S&P with its count of fresh one-year lows falling below its count of fresh highs would mean the narrowness flagged in these pages was an artifact. Issue 195's five-session clock is void, and the reason is that it was written badly. It required Materials, Consumer Staples and Utilities all to return to red while Technology held above plus two hundred. On the very first session Technology fell to 193.5 and Utilities went red, so the two halves moved in opposite directions at once and the test can no longer be satisfied either way. A test whose halves can move against each other was never a test, and it will not be restated here in a shape that lets it pass. Today's condition is single-sided and checkable: if the ten-year closes below 5.00% within five completed sessions and Technology returns to a green verdict inside that same window, then Wednesday was a rate shock that these pages read as a change of regime, and the correct response is to print that sentence here. The market-risk gauge gave up the green it had held for two bars, and the S&P Risk study on the Golden Terminal carries the full history of that gauge instead of only today's reading.

The Daily Dashboard reads the same tape on a longer clock, tracking where capital is being committed across the full cycle instead of where it moved in a single session. A fifteen basis point day in the ten-year is noise on a quarterly chart and a regime question on a monthly one, and the difference between those two readings is the whole argument this week.

If today's read sharpened your thinking, send it to one person who would rather see the arithmetic than the narrative. Readers who bring three new subscribers to the free list receive thirty percent off any paid tier.

Every directional and magnitude claim above, checked against the live tape. No “trust me, bro”, these are the numbers that pay for your subscription.

The CCI(20) engine recomputes from completed bars every run and its prior-session output must reproduce the previous issue's published current values exactly, or the run is not trusted. This run reproduced Issue 195 on 15 of 15 instruments: XLK +237.3, XLC +113.0, XLY -47.0, XLP -102.0, XLV -9.8, XLF -173.1, XLI -69.9, XLE -133.5, XLB -81.7, XLU -154.6, XLRE -111.0, SPY +135.6, plus the stock-level readings COST -69.7, MU +190.3 and MSFT +19.6. No drift.

Confirmed on two or more independent feeds and stated as fact above: the ten-year Treasury yield reaching a 19-year high on Wednesday, its highest since July 2007, in its biggest one-day move in nearly eighteen months (CNBC, CNBC International TV, WSJ, corroborated by our own rate series); the September S&P Global flash PMIs printing far above estimate and prior on all three lines (FMP economics-calendar actuals plus Seeking Alpha and WSJ); the five-year note auction stopping at 5.033% against 4.393% (FMP economics-calendar plus Barron's); New York Fed President Williams saying another rate increase this year is a reasonable expectation and that the Fed is done with explicit forward guidance (CNBC, Barron's); the United States and China extending their trade truce by two months to January 10 as Xi Jinping arrived in Washington for his first state visit in eleven years (CNBC, Bloomberg); McDonald's holding an analyst and investor day and laying out a ten-year plan including roughly $8.5 billion of restaurant investment (posted transcript plus independent coverage); Meta's Connect announcements including the Muse assistant expansion, a handheld device, $1,299 virtual-reality glasses and camera-free audio glasses (CNBC, Reuters, NYT, TechCrunch, Business Insider, The Guardian); global debt reaching a record above $365 trillion per the Institute of International Finance (CNBC, CNBC International TV); and crude rising about four percent on Wednesday before easing Thursday morning as Iran said it remains open to diplomacy (Reuters plus our own tape).

Hedged as single-source and not stated as fact: the characterization of the five-year auction as disastrous (Barron's alone; the 64bp jump in the stop rate is the hard number and is the version used above); the Energy Secretary's Thursday narrowing from a diesel export ban to export restrictions (WSJ alone, and flagged as one feed in the copy); the GDPNow 5.1% third-quarter estimate and the five-percent-GDP read on the flash surveys (Seeking Alpha alone); Japan's ten-year yield hitting a thirty-year high (CNBC alone); a price-target increase on Meta tied to the Muse agent (MarketWatch alone); profit-taking as the cause of Meta's premarket decline (Barron's alone); Norway raising rates with Sweden signalling a move before year end (Reuters alone); and Amazon blocking Meta's shopping agent, where two outlets trace to one underlying statement and are treated as one.

No cause stated: the Alphabet, Oracle, Broadcom, NextEra and Home Depot session declines, and the premarket offers across the chip complex, are reported as tape after multiple feeds failed to confirm a same-day catalyst. Withheld: market-implied rate-probability figures, consistent with every issue since 187; and per the competitive-read limits, the outside desks' forward inflation and yield targets, their single-desk miner margin and buyback figures, their household-affordability citations, and their political prediction-market odds.

No new weekly fifty-two-week extremes screen published by the outside desk this run; that publication runs on a Sunday cadence. The divergence recorded on 9/20 remains carried forward unresolved and is deliberately not re-graded.

Crude oil, gold, silver, broad commodity and dollar exposures are expressed through exchange-traded fund proxies (USO, GLD, SLV, DBC, UUP) because futures contracts are not entitled on the current data plan. Proxy returns include fund structure effects and do not equal spot returns. Index exposures use SPY, QQQ, DIA and IWM as proxies for the S&P 500, Nasdaq, Dow and Russell 2000. The index-drift overlay uses the S&P premarket proxy because the futures snapshot endpoint returns an authorization error and was not called.

The standing note was written instead of skipped this run. It is the final sentence of the Final Word, tied to the market-risk gauge losing the green it had held for two bars, and points to the S&P Risk study. It states only whitelisted facts, carries no performance claim, no recommendation and no roster count, and appears once in this issue.

Disclaimer. The Daily Update is a general-circulation editorial publication of impersonal commentary and does not provide personalized investment advice. Nothing in this issue is a recommendation to buy or sell any security, and no statement here is tailored to the financial situation, objectives or needs of any individual reader. Market data is drawn from completed daily bars and third-party feeds and is believed accurate but is not guaranteed. Estimates, price targets and analyst consensus figures are third-party sell-side data, are labeled as such where used, and are not forecasts of this publication. Past performance does not indicate future results. Readers should consult a licensed professional before acting on anything they read anywhere, including here.

The Daily Update · Issue 196 · Volume III · Filed from Taintsville, Florida · Copyright 2026 Golden Terminal. All rights reserved.