ISSUE #16  ·  OCTOBER 9, 2026
THE FRIDAY FIVE
The free weekly read on market conditions, sector rotation, and stock ideas.

What's going on, everybody?

I hope you all had a great week. This is Friday Five, Issue #16, and this was the week the market paid for signatures and stopped paying for projections.

For four sessions it was an ordinary technology week. The technology sector fund was up 1.81 percent through Wednesday against 1.77 percent for the S&P 500, the same narrow leadership that has held since late August. Then on Thursday the Financial Times reported that OpenAI's annualized revenue was near 50 billion dollars rather than the 68 billion that investors had been citing, and the difference turned out to be a counting convention: the larger number included revenue earned through partners on a gross basis.

No customer left. No contract changed. The same investor materials showed the business growing 77 percent on a run rate basis, with the enterprise segment up 107 percent. Roughly 20 billion dollars of apparent revenue simply stopped existing, and the complex around it fell between 3 and 8 percent in one session. CoreWeave lost 7.77 percent that day, Arm 6.48 percent, Oracle 5.48 percent, Nvidia 2.94 percent.

Underneath that, the week had a defensive shape that had nothing to do with artificial intelligence. On October 2 the September employment report came in at 29,000 payrolls against a consensus near 84,000, unemployment rose to 4.2 percent, July and August were revised down by a combined 60,000, and annual wage growth of 3.0 percent was the slowest since 2021. Energy led the week at 4.05 percent, consumer staples at 3.85 and utilities at 3.50. A weak labour market and crude carrying a geopolitical premium is a hard combination to own growth into.

And the five names in this issue all went the other way, because each of them was holding a number somebody had signed: two all cash offers at 205 and 32.05 dollars a share, a 20 year power contract for 890 megawatts, a reported quarter, and a government inventory reading at a four decade low. That is the whole issue. Alright, let's get into it.

The Week in 60 Seconds

01The S&P 500 rose 1.29 percent, the Nasdaq 1.20 percent, and the Russell 2000 fell 0.45 percent, so the broadening never reached smaller companies.
02September payrolls came in at 29,000 against a consensus near 84,000, unemployment rose to 4.2 percent, and wage growth was the slowest since 2021.
03Energy led at 4.05 percent, staples at 3.85 and utilities at 3.50, while technology finished flat and industrials last; that is a defensive mix.
04On October 8 alone technology fell 1.79 percent while the equal weight index rose 1.21, a three point reversal on one report about one private company.
05Two of this week's five are all cash takeover targets, so what is left in them is 5.8 and 3.2 percent, not another thirty.

This Week in One Minute

The market paid for signed numbers and stopped paying for projected ones.

This was an up week with a strange shape. The S&P 500 rose 1.29 percent and the Nasdaq Composite 1.20 percent, but the Russell 2000 fell 0.45 percent, and the leadership was energy at 4.05 percent, consumer staples at 3.85 percent and utilities at 3.50 percent, with technology flat and industrials last. That is a defensive, inflation hedged mix, and the September employment report on October 2 explains why: payrolls came in at 29,000 against a consensus near 84,000, unemployment rose to 4.2 percent, July and August were revised down by a combined 60,000, and annual wage growth of 3.0 percent was the slowest since 2021. A weak labour market and crude carrying a geopolitical premium is a difficult combination to own growth into. Then there is October 8, which is really the whole week. For four sessions technology led, up 1.81 percent through Wednesday. On Thursday the Financial Times reported that OpenAI's annualized revenue was near 50 billion dollars rather than the 68 billion widely cited, a difference created by whether partner revenue is counted gross, with the same materials showing run rate growth of 77 percent. Technology fell 1.79 percent that session and the equal weight index rose 1.21 percent. Meanwhile the week's actual winners all held a number somebody had signed: two cash offers, a 20 year power contract, a reported quarter and a diesel inventory reading at a four decade low.

  • Nothing in this issue moved on a theme. Two all cash acquisition agreements, one 20 year power purchase agreement, one reported quarter and one Energy Information Administration inventory reading drove all five names, and every figure traces to a filing, a company announcement or a government release.
  • Breadth improved for the first time in seven weeks and I do not yet believe it. The equal weight index beat the cap weight index by 0.07 percentage points, but the entire margin was earned on October 8 when the largest companies were sold, and the Russell 2000 fell 0.45 percent over the same window.
  • Two of the five have almost nothing left to give. PTC jumped 33.5 percent to within 5.8 percent of a 205 dollar cash offer and Option Care Health jumped 32.6 percent to within 3.2 percent of a 32.05 dollar one, which is the point of including them.

The Week Ahead

WHAT I EXPECT  ·  64%  ·  MODERATE

I expect the market to keep paying for signed numbers and to keep discounting projected ones, and the cleanest place to test that is Constellation Energy. On October 6 it signed a 20 year agreement to put 890 megawatts of new nuclear capacity on the PJM grid for Google, and the stock rose 12.25 percent that day. On October 8 it fell 4.85 percent, giving back about 40 percent of the gain, because the Financial Times reported that a private company it does not contract with had been counting its revenue differently than investors assumed. Nothing in the agreement changed. The 68 billion dollar figure became 50 billion because the larger number counted partner revenue gross, and the same materials showed run rate growth of 77 percent. So from the October 9 close to the October 15 close I expect Constellation Energy to beat the utility sector fund XLU, because a contracted cash flow does not reprice on another company's accounting convention. I put this at 64 percent rather than higher for two reasons, and the first one is about me. My own record says I have been too bullish on three of the last four misses, and every one of those misses extrapolated the previous week's winner. This call runs the other way, against the newest move, which is a different mistake to make and not an immunity from making one. The second reason is September CPI at 8:30 in the morning on October 14, inside this window, which can settle the rate question against every rate sensitive name in the sector including this one.

What would prove me wrong. Constellation Energy trails the utility sector fund XLU by more than 3.0 percentage points from the October 9 close to the October 15 close, measured as the percentage change in CEG minus the percentage change in XLU. That threshold sits 0.62 of one standard deviation from where the spread is today, so an ordinary week can reach it. A random walk carrying this spread's own measured volatility would trigger it 53 percent of the time, which is the bar this call has to clear to mean anything.

Last week's call, graded. It trailed by 2.04 points, so the call failed by half a point past its own line. The path is the uncomfortable part. Through the Wednesday October 7 close the call was ahead by 0.38 points and would have been recorded as a survival. The October 8 session alone moved the spread 2.41 points against me, with XLK down 1.79 percent while RSP rose 1.21 percent. The mechanism was wrong too, not just the direction, and that is the part I want on the record. My whole argument rested on a rising term premium doing the sorting. On October 8 the 30 year Treasury yield FELL 5.5 basis points and technology had its worst session of the week anyway. The discount rate was not what sorted this market. A report that a private company had been counting its revenue differently was. There is a second correction to disclose. The grading system first reported this call as having survived, because the data source returned the October 8 close for exchange traded funds as empty and the check silently measured only through October 7. I caught it, regraded the call by hand against the complete week, and logged the fault so it cannot recur. Grading my own forecast on four of its five sessions and reporting a survival would have been worse than being wrong.

By how much. I said I was wrong if technology trailed the average large cap by more than 1.5 points. It trailed by 2.04, so there was no ambiguity by the end. But it came to the line from the right side: with one session left the call was ahead by 0.38 points, and a single day moved it 2.41 points. Three of my last four misses now carry the same direction, too bullish, and all three extrapolated whatever had won the week before. That is a pattern in me rather than in the market, and this week's call deliberately runs the other way.

Dates that matter.

The biggest risk to all of this. October 8 was information rather than noise, and the market has started repricing artificial intelligence capital spending itself rather than one private company's revenue line. In that world a signed 20 year power agreement offers less protection than it looks like, because what gets marked down is the 2028 demand that justifies building 890 megawatts of new capacity, not the 2026 signature that sold it. The same logic reaches the electrical and data centre equipment makers that made industrials the worst sector of the week.

Why I think so. The complex fell on positioning, not on cash flow. CoreWeave lost 7.89 percent over the window, Arm 5.83 percent, Micron 5.61 percent and Oracle 1.72 percent, in a week the S&P 500 rose 1.29 percent, and almost all of the damage landed in the October 8 session alone. A group that gives up between 3 and 8 percent in a day because a customer restated how it counts partner revenue is carrying crowded positioning rather than a valuation problem. Industrials finished last at minus 0.14 percent with Eaton down 2.92 percent and Caterpillar down 3.65 percent, which is the same trade wearing different clothes.

What would confirm it. The technology sector fund XLK closing below 190 on or before October 22, 2026. It closed at 197.78 on October 8, so that is a further 3.94 percent decline, about 0.8 standard deviations of a nine session move in this observable. I put that at 30 percent.

Market Risk Score

61 out of 100. Elevated. Down 3 points from last week.

Breadth improved, nine of eleven sectors rose and the VIX fell 5.98 percent to 15.41 from 16.39. There was no fear in this tape at any point, including on Thursday, and all of that argues for a lower reading. Three things hold it up. The labour market deteriorated materially. Crude is carrying a geopolitical premium rather than a demand one, with roughly a quarter of Gulf production shut in. And a single headline removed between 3 and 8 percent from major infrastructure names in one session, which measures how crowded the positioning is rather than how expensive the stocks are.

One disclosure about this number. The calibration ledger now has six matured readings of the Market Risk Score against what the market actually did next, and the relationship is still backwards: higher readings have preceded stronger forward markets, with a correlation of 0.422. Six readings is far too few to conclude anything, but it is enough that I will not let you read this score as a forward signal. It describes the conditions I can measure today. I would rather keep saying that than quietly print a number that implies otherwise.

Sector Rotation Update

Into: Energy · Consumer Staples · Utilities · Consumer Discretionary
Out of: Industrials · Technology · Real Estate · Health Care

Is the rotation durable? Not yet, and I want to be precise about why. The equal weight index beat the cap weight index by 0.07 percentage points over the week, the first positive reading in seven, and every bit of that margin was earned on Thursday. The tell against durability is the Russell 2000, which fell 0.45 percent over the same window. A genuine broadening lifts the smaller companies first, and this one never reached them. What happened was de risking inside large caps, into energy, staples and utilities. Two honest caveats on the labels. I have retired the Improving label from this issue: the calibration record has it right 9 percent of the time at a one month horizon across twelve readings, which makes it worse than useless. And the rotation read overall scores 59 percent at one month across 50 matured readings, so I lean on the direction and not on the labels.

Breadth: What Actually Improved

This needs its own section this week, because the headline number says breadth improved and the way it improved matters more than the fact.

The equal weight S&P 500 fund rose 1.37 percent against 1.30 percent for the cap weight fund, a spread of plus 0.07 percentage points. That is the first positive reading in seven weeks. The six before it ran minus 0.58, minus 0.90, minus 1.14, minus 0.56, minus 2.03 and minus 0.18, an average of minus 0.90, so the reading has moved 0.97 points against its own recent average. The sign flipped.

It flipped from the top down. On October 8 alone the equal weight fund rose 1.21 percent while the cap weight fund fell 0.42 percent, a 1.63 point spread in a single session, and the sellers were named: CoreWeave down 7.77 percent, Arm 6.48, Oracle 5.48, Super Micro 4.83, Micron 4.79, Broadcom 4.35, AMD 3.90 and Nvidia 2.94. The technology sector fund fell 1.79 percent. Breadth did not improve because the average company got better. It improved because the largest companies got sold.

That distinction decides what the reading is worth. A broadening driven by the median company re rating is durable, because it reflects earnings reaching more of the index. A spread created by the top of the index falling is arithmetic, and it reverses the moment the top stops falling.

The Russell 2000 settles which one this was. It fell 0.45 percent over the same window, and 1.37 percent measured from last Friday's close. Breadth improved inside the large cap index and did not improve at all below it. There are two readings that would settle it properly inside the next two weeks: the Russell 2000 closing above 2880, which would say the broadening is real, and September consumer price index on October 14, which decides whether the inflation hedged half of this week's leadership was reading something true.

The Five

01  ·  ACQUISITION

PTC Inc. (PTC) +32.27%

Technology / Industrial Software  ·  Friday Five Score 84  ·  News Spike

On October 4 PTC signed an agreement to be acquired by Schneider Electric for 205 dollars a share in cash. The filing landed on October 5 and the stock went from 144.03 to 192.26 in one session, up 33.5 percent, then spent the next three sessions doing nothing at all: 193.00, 193.60, 193.73. That flatness is the most useful thing here. A stock pinned inside a dollar and a half for three days is not a market forming an opinion, it is a register that has changed hands to people who care about one thing, whether the deal closes. At 193.73 against a 205 dollar offer, what is left is 5.82 percent, and it is capped there, to a close the company anticipates by the third quarter of 2027. This is in the issue precisely because the 33 percent is gone. It is the clearest lesson of the week in what a signed number does to a share price, and in why a big move is not the same thing as an opportunity.

Why it made the five. Three things put this ahead of the rest of the screen. The cause is not in dispute: a regulatory filing, a stated price and a single session. The move is also almost entirely company specific, because technology as a sector finished the week flat at minus 0.02 percent, so none of this is a sector tailwind. And it is the one pick that teaches something a reader can use every week rather than only this week. Most people see a 33 percent gap and feel they have missed something. The arithmetic says the opposite: the gap is the market pricing the offer in full, and the remainder is the fee for carrying deal risk for eleven months.

The case against it. The asymmetry is the whole bear case and it is severe. The upside is 5.82 percent, fixed. The downside, if the deal breaks, is a return toward the 144.03 close of October 2, roughly 26 percent below where the stock sits now. So a reader is risking about four and a half points of loss for every point of gain, on an outcome decided by regulators and a shareholder vote rather than by the business. The deal is also large and cross border, at 22.6 billion dollars, which invites review in more than one jurisdiction, and the timeline runs to the third quarter of 2027, which is a long time for anything to go wrong. The 700 million dollar termination fee protects Schneider Electric, not the shareholder.

WHAT TO WATCH NEXT

A shareholder vote at a special meeting, and antitrust clearance in more than one jurisdiction. No meeting date has been set, and the company anticipates closing by the third quarter of 2027

Confirms the thesis: Clearances progress and the spread to 205 dollars narrows toward the cost of waiting

Breaks it: A regulator opens an in depth review, which widens the spread and puts the 144.03 pre announcement price back in play

Source: U.S. Securities and Exchange Commission, PTC Inc. Form 8-K reporting the Agreement and Plan of Merger with Schneider Electric SE, October 5, 2026. https://www.sec.gov/Archives/edgar/data/0000857005/000119312526413124/d174191d8k.htm

02  ·  ACQUISITION

Option Care Health, Inc. (OPCH) +37.51%

Health Care / Home Infusion Services  ·  Friday Five Score 82  ·  News Spike

The same structure as PTC, one day later and with tighter numbers, which is what makes the pair worth reading together. On October 6 Option Care Health agreed to be acquired for 32.05 dollars a share in cash by Clayton Dubilier and Rice, holding about 51 percent, alongside McKesson at about 49 percent, at an enterprise value near 5.8 billion dollars. The stock went from 23.37 to 31.00 in one session, up 32.6 percent, then printed 31.01 and 31.05. At 31.05 the remaining return is 3.22 percent, to a close expected in the first half of 2027. Compare that with PTC: 5.82 percent to a later close. The wider spread is not the better deal. It is the market charging more for a bigger, slower transaction with more regulators in it, and seeing those two numbers side by side is the fastest way to understand how deal risk actually gets priced.

Why it made the five. It earns its place for the contrast rather than for the return. Health Care rose 1.18 percent on the week and lagged the index, so this is a 37.51 percent move with no sector help at all, which is exactly the profile the screen is built to find. It also gives the issue a second, independent reading on the same question. One cash deal tells you a price. Two cash deals with different spreads and different closing dates tell you what the market thinks the risks are worth, and that comparison is information a single name cannot give.

The case against it. Worse asymmetry than PTC: 3.22 percent of upside against roughly 25 percent of downside to the 23.37 close of October 5. There is also a risk here that PTC does not carry. A private equity led buyout depends on financing that has to be raised, where a corporate acquirer pays from its own balance sheet, so a sharp tightening in credit is a direct threat to this deal and not merely a market event. And the regulatory question is pointed rather than routine: McKesson is already one of the largest pharmaceutical distributors in the country, and taking 49 percent of a home infusion network operating in all 50 states is the kind of vertical combination that attracts a closer look. A reader should also notice that the company will remain separately managed, which is the structure, not a reassurance about the outcome.

WHAT TO WATCH NEXT

A shareholder vote and regulatory clearances, with closing expected in the first half of 2027. No meeting date has been set, and third quarter results are expected late October though the company has not announced a date

Confirms the thesis: Clearance proceeds without a second request and the spread grinds toward the 32.05 dollar offer

Breaks it: A financing problem at the private equity buyer or a regulator objecting to a large distributor taking 49 percent of a national infusion network

Source: U.S. Securities and Exchange Commission, Option Care Health, Inc. Form 8-K, exhibit 99.1, October 6, 2026. https://www.sec.gov/Archives/edgar/data/0001014739/000110465926113904/tm2627097d1_ex99-1.htm

03  ·  MARGIN EXPANSION

Marathon Petroleum Corporation (MPC) +10.28%

Energy / Refining  ·  Friday Five Score 81  ·  Extended

The one pick in this issue whose number nobody signed. Marathon Petroleum rose 10.28 percent on the week, and 4.77 percent of it on October 8, on a refining margin rather than an agreement. The Energy Information Administration reported distillate inventories near 105.1 million barrels for the week ending October 2, the lowest level for early October in a series that starts in the early 1980s, with refineries already running at 92.7 percent of capacity and September retail diesel averaging 6.29 dollars a gallon. Then Hurricane Isaias shut in roughly 512,000 barrels a day of Gulf production, about a quarter of the offshore total. The detail that matters is that crude itself did not drive this: the crude fund USO fell 1.63 percent over the same window. This was about what refiners earn turning oil into diesel, in a market with no inventory cushion left.

Why it made the five. It is the only name in the issue with an uncapped outcome, which is why it is here alongside two deal spreads that cannot go up. The move also survives the test the screen cares about most: Energy rose 4.05 percent as a sector, so roughly six points of this are specific to refining rather than to the oil price, and the crude fund falling over the same week rules out the simple explanation. Low inventories, high utilization and a supply interruption is the clearest version of a margin story there is, and the data behind it comes from a government release rather than from anybody's forecast.

The case against it. This is the sharpest risk in the issue and it resolves almost immediately. Marathon's Garyville refinery, at 617,000 barrels a day the largest plant named in the storm's projected path, sits in the track of a hurricane forecast to make landfall the night this issue is delivered. The market has spent the week paying for the margin that refinery outages create and discounting the chance that this particular company owns one of the refineries that goes down. That can be right, because an outage at a competitor widens the spread further. It can also be badly wrong. A 10.28 percent move ahead of a landfall is not a de risked position, it is a bet with a deadline, and a distillate rebuild once the storm passes would take the margin back as quickly as the storm gave it.

WHAT TO WATCH NEXT

Nov 3: Third quarter results and an 11 a.m. Eastern call on November 3, announced by the company. Before that, Hurricane Isaias is forecast to make landfall the night of October 9

Confirms the thesis: Garyville comes through the storm intact and the quarter captures the wide margins the inventory data implies

Breaks it: Storm damage at Garyville, or a distillate rebuild that narrows the margin before the quarter is in the bank

Source: U.S. Energy Information Administration, Weekly Petroleum Status Report highlights, week ending October 2, 2026, October 7, 2026. https://www.eia.gov/petroleum/supply/weekly/pdf/highlights.pdf

04  ·  CONTRACT WIN

Constellation Energy Corporation (CEG) +10.10%

Utilities / Nuclear Generation  ·  Friday Five Score 86  ·  Watch Pullback

This is the name that explains the week. On October 6 Constellation Energy and Google announced a 20 year power purchase agreement bringing 890 megawatts of new nuclear capacity onto the PJM grid, built through uprates at 11 of Constellation's reactors in Illinois, Pennsylvania and New Jersey, with Constellation committing more than 4.3 billion dollars of its own capital and first delivery expected in 2028. A separate 15 year agreement covers another 2,700 megawatts from the existing fleet. It is the largest uprate deal ever struck between a power producer and a technology company, and the stock rose 12.25 percent that day. Two sessions later it fell 4.85 percent, giving back about 40 percent of that gain, because the Financial Times reported that a different company, one that is not a counterparty to any of this, had been counting its revenue differently than investors assumed. Not one word of the agreement changed.

Why it made the five. It made the five because it is the cleanest natural experiment the week offered. A company signed a contract on Tuesday and lost a fifth of the resulting gain on Thursday on news about somebody else's accounting, which separates what a business earns from what a theme is worth more precisely than any argument could. Utilities rose 3.50 percent as a sector, so about six and a half points of this are the agreement rather than the sector. It is also the name my forward call this week is built on, which means a reader can check my reasoning against a published number rather than take it on trust.

The case against it. Two things cut against it, and the second is the serious one. The 4.3 billion dollars is Constellation's own capital commitment, money it spends, not money it receives, and the capacity does not arrive until 2028, so a reader buying now is paying today for generation that does not exist yet against a contract that does. The deeper problem is that October 8 proved this stock trades on sentiment about artificial intelligence spending that the company cannot control. If the market is beginning to reprice that spending in general rather than one company's revenue line, a signed contract offers less protection than it looks like, because what gets marked down is the 2028 demand that justifies the build, not the 2026 signature that sold it. Sector labels hid this all week: utilities looked defensive and its data centre power names fell between 2 and 6 percent in a single session.

WHAT TO WATCH NEXT

Third quarter results expected in early November, though the company has not announced a date. Before that, September CPI on October 14 is the nearer test for every rate sensitive name in the sector

Confirms the thesis: The stock holds above its 267.62 close of October 5, meaning the market still prices the agreement at something

Breaks it: A slide back through the pre announcement level, which would say the market has decided contracted 2028 capacity is worth nothing if the demand behind it is in question

Source: Google, Google and Constellation Announce Landmark Agreement to Bring 890 MW of New Nuclear Capacity to PJM Grid as Part of Long-Term Power Deal, October 6, 2026. https://www.googlecloudpresscorner.com/2026-10-06-Google-and-Constellation-Announce-Landmark-Agreement-to-Bring-890-MW-of-New-Nuclear-Capacity-to-PJM-Grid-as-Part-of-Long-Term-Power-Deal

05  ·  EARNINGS BEAT

Constellation Brands, Inc. (STZ) +9.42%

Consumer Staples / Beverages  ·  Friday Five Score 76  ·  Watch Pullback

The only earnings driven name in the issue, and the one with a catch worth learning. Constellation Brands reported second quarter fiscal 2027 results after the close on October 6: net sales of 2,633 million dollars, up 6 percent, comparable earnings of 3.74 dollars a share against a 3.56 dollar consensus, beer net sales up about 5 percent and beer shipment volumes up 5.5 percent, with 400 million dollars returned through buybacks and dividends. The stock rose 2.35 percent on October 7 and another 4.43 percent on October 8, finishing the week up 9.42 percent. The catch is underneath the beat. Depletions, which measure what distributors actually sell on to retailers, fell 0.6 percent, with Modelo Especial down about 2 percent and Corona Extra down about 5 percent. Shipments growing 5.5 percent while depletions fall means the quarter put beer into the distribution channel rather than into anybody's fridge.

Why it made the five. It belongs in the five because it is the week's leadership in one company. Consumer staples was the second best sector at 3.85 percent, which almost never happens in a week the index makes a new high, and this is the staples name with an actual dated event behind the move rather than a defensive bid. Roughly five and a half points of the 9.42 percent are company specific. It also sits directly across from the September labour data: a beverage company gaining dollar share while management calls the consumer stretched is the most informative kind of staples result there is.

The case against it. The market has priced the depletion number as noise, and it may not be. A 9.42 percent move on a 5 percent earnings beat says investors read the shipment growth as demand, when the two largest brands in the portfolio went backwards at the point of sale. Channel inventory is borrowed revenue: it has to be sold through eventually, and if it is not, the next quarter absorbs it. Margin is the second warning. Comparable operating income rose 1 percent against net sales up 6 percent, so the growth cost something to buy. And management's own description of a stretched United States consumer sits badly with a week in which September payrolls came in at 29,000 and unemployment rose to 4.2 percent. There was also no guidance raise alongside the beat, which is a weaker configuration than the headline suggests.

WHAT TO WATCH NEXT

Third quarter fiscal 2027 results expected in early January, though the company has not announced a date. The nearer reading is the next distributor depletion update, which is where a channel build shows up first

Confirms the thesis: Depletions turn positive and shipments stay close to them, which would make this quarter a timing effect

Breaks it: Depletions fall again while shipments slow to match, which turns the channel inventory into a drag on the next two quarters

Source: Constellation Brands, Inc., Constellation Brands Reports Second Quarter Fiscal 2027 Financial Results, October 6, 2026. https://ir.cbrands.com/news-events/press-releases/detail/346/constellation-brands-reports-second-quarter-fiscal-2027-financial-results

Stock of the Week

Constellation Energy (CEG)

Constellation Energy is the Stock of the Week because it is the week's argument compressed into one chart. On Tuesday October 6 it announced the largest nuclear uprate agreement ever struck with a technology company: 890 megawatts of new capacity onto the PJM grid for Google, under a 20 year contract, plus a separate 15 year agreement for another 2,700 megawatts from the existing fleet. The stock rose 12.25 percent. On Thursday October 8 it fell 4.85 percent, surrendering about 40 percent of that gain, and the reason had nothing to do with Constellation Energy, Google, PJM, or any of the 11 reactors involved. The Financial Times reported that OpenAI's annualized revenue was near 50 billion dollars rather than the 68 billion that investors had been citing, and the gap turned out to be a counting convention: the larger figure included partner revenue gross, and the same investor materials showed the business growing 77 percent on a run rate basis with enterprise up 107 percent. No customer left. No contract changed. A measurement convention was corrected, and roughly 20 billion dollars of apparent revenue stopped existing. What that session showed is where the risk in this market actually sits. It is not in the signed agreements, which are specific, dated and enforceable. It is in the projected demand that justifies signing them. A 20 year power contract is as close to a certain cash flow as a utility gets, and it still lost a fifth of its announcement value in one day because the market had a moment of doubt about a number that was never audited in the first place. The useful habit is to ask, of anything that moves, whether the number behind it was signed or merely estimated, and then to notice that the market does not always price the difference consistently. This week it paid for signatures on Monday and Tuesday, and on Thursday it took its doubts out on a signature anyway.

What Could Change My Mind?

A hot September inflation print on October 14. This is the one. My forward call rests on a contracted utility beating its own sector, and a hot consumer price index lands directly on every rate sensitive name in that sector, including this one. It would also validate the inflation hedged leadership of energy and staples, which would mean the week's defensive shape was reading something real rather than reacting to one labour print. I put a hot print doing real damage to this call at about 30 percent.

The technology sector fund below 190 by October 22. If the market is repricing artificial intelligence capital spending in general rather than one company's revenue line, then a signed contract protects less than it looks like, because what gets marked down is the 2028 demand that justifies building the capacity. That is the published risk in this issue and the level is the confirming reading. The same logic reaches the electrical and data centre equipment makers that made industrials the worst sector of the week.

The Russell 2000 above 2880. That would be a genuine return of risk appetite below the large cap index, and it would mean my durability verdict on this week's rotation was wrong. I have said the broadening was arithmetic rather than real. A Russell breakout is the cleanest way for the market to tell me otherwise.

A direct hit at Garyville. Marathon Petroleum rose 10.28 percent on a refining margin, and its own largest refinery sits in the projected path of a hurricane forecast to land the night this issue reaches you. If that plant takes damage, one of the five picks stops being a story about margins and becomes a story about repairs, inside 48 hours of publication.

Scoreboard

I publish the losers beside the winners, and I publish the measurement that flatters me least first.

Last week's five, one week on. Measured from the October 2 close to the October 8 close, against the S&P 500 fund at plus 0.56 percent:

Four of the five beat the index and the group averaged plus 3.63 percent against plus 0.56 percent for the S&P 500 fund, so 3.07 points of relative gain. United Therapeutics is the loser on the table at minus 1.65 percent, and it was the largest issue week mover of the five at plus 21 percent, which is the pattern worth noticing: the biggest move in the issue week was the one that gave the most back.

A good week for the picks and a bad one for the call. In the same five sessions that those five averaged 3.07 points of relative gain, the forward market call in that issue was falsified. Stock selection worked and the market read did not. Those are two separate skills and this Scoreboard has now measured them separately fifteen times, which is the only reason I can say that with a number instead of a feeling.

The score column, still not working. Across 55 matured picks the numeric score is still only weakly related to what happens next. Picks scored 85 and above have averaged plus 5.11 percent against the index at one month, and picks scored below 75 have averaged plus 7.56 percent. The low scores are still beating the high scores. I keep publishing the score because hiding a number that is not working would be worse, but read the catalyst and the case against before you read the score.

The full record, two ways. Measured in the week each pick appeared, 75 picks across 15 issues have returned an average of 12.97 percent against 0.16 percent for the benchmark. That number flatters me and I will tell you exactly why: a stock is selected because it already moved that week, so the issue week return is partly a measurement of the selection rather than of the idea. The honest number starts the day you could have acted. Measured from publication forward one week, the 70 picks whose week has actually finished average 0.15 points ahead of the benchmark with a 52.9 percent beat rate, which is a coin flip. Last week's five are not in that number yet: their week closes on the October 9 session, after this issue is delivered, so counting them now would be measuring four sessions and calling it five. At one month, the 55 picks that have matured average 2.38 points ahead with a 61.8 percent beat rate. One week is noise and one month is not, and that gap is the most useful thing this record has told me so far.

The screens, graded. I reject far more names than I publish and those rejections are tracked too. Across 203 logged rejections, 73 percent of the names I screened out went on to underperform at one month: 81 correctly avoided, 22 neutral, and 30 that were real winners I passed on. The most reliable screen remains the catalyst check, where 100 percent of the names rejected for having no dated catalyst went on to lag. That screen did most of the work this week: it removed Walmart at plus 6.04 percent, NRG at plus 9.69, HCA at plus 5.11 and Elevance at plus 5.05, all strong weeks with no company announcement I could point to.

Terms in This Issue

Annualized run rate. Taking the most recent period's revenue and multiplying it out to a full year, as though nothing changed. It is a projection dressed as a current figure, and because there is no single rule for what to include, two companies can describe the same business with very different numbers.

Power purchase agreement. A long term contract in which a buyer agrees to take a set amount of electricity at an agreed price for a fixed number of years. For the generator it turns an uncertain future price into a known one, which is why these contracts change how a utility is valued.

Crack spread. The gap between what a refiner pays for crude oil and what it gets for the gasoline and diesel it makes from it. Refiner profits move with this gap, not with the oil price, which is why a refiner can rise in a week crude falls.

Depletions. In the drinks business, the volume that distributors sell on to bars and shops, as opposed to shipments, which is what the producer sells to the distributors. When shipments grow faster than depletions, stock is piling up in the middle and the producer has borrowed from a future quarter.

Termination fee. A sum written into a takeover agreement that one side pays the other if it walks away under defined circumstances. It is usually a small percentage of the deal, and it protects the parties to the agreement rather than the shareholders who bought in after it was announced.

A Possible Next Step

I'm testing interest in a future Edge Report with entry zones, invalidation levels, sizing tiers, and exit frameworks. Nothing is for sale and there is nothing to subscribe to yet. I am genuinely trying to work out whether that is something readers of this letter want, so if it is, tell me and I will build it.

REGISTER INTEREST →

And if this was useful, the quickest thing you can do is send it to one person who would get something out of it. That is how this grows.

The Friday Five is for informational and educational purposes only. It is not investment advice, and it is not a recommendation to buy or sell any security. Past performance does not predict future results. All figures are measured from closing prices on the dates stated. Do your own research.