What's going on, everybody?
I hope you all had a great week. This is Friday Five, Issue #15, and this was the week the market stopped paying for stories.
On Wednesday the August inflation report came in cooler than anyone expected, 3.4 percent against a 3.7 percent forecast, and the core reading at 3.0 percent against 3.3. Traders immediately cut the odds of an October rate increase to about 37 percent from roughly a coin flip the day before, and the three month Treasury bill yield fell nine basis points across the week.
The 30 year Treasury yield went up anyway. It closed at 5.638 percent on September 30, the highest close of this cycle, and finished the week at 5.603, up 14.2 basis points. Short rates down nine, long rates up fourteen. That is a 23 basis point steepening in five sessions, and it happened on good inflation news.
That tells you something specific. A long bond that will not rally when inflation surprises to the downside is not being sold because of inflation. It is being sold because of supply, and there are two Treasury auctions next week that will test exactly that.
The equity market took the hit broadly. Eight of eleven sectors fell against three the week before. The S&P 500 fell 0.49 percent, the Russell 2000 fell 1.02 percent, and the equal weight index fell more than the cap weight one. And then five companies went up anyway, by between 7.7 and 21 percent, every single one of them on a dated disclosure that landed inside the same five sessions. That is the whole issue. Alright, let's get into it.
The Week in 60 Seconds
| 01 | The S&P 500 fell 0.49 percent, the Russell 2000 fell 1.02 percent, and eight of eleven sectors finished lower. |
| 02 | The August inflation print came in cooler than expected on September 30 and the 30 year Treasury yield still closed at 5.638 percent, the highest of the cycle. |
| 03 | The three month bill yield fell nine basis points while the 30 year rose 14.2, a 23 basis point steepening in a single week. |
| 04 | Accenture rose 19.67 percent on record bookings, United Therapeutics 21.00 percent on a patent ruling, and Carnival 15.05 percent on a record quarter. |
| 05 | The risk I named last week fired eleven days early; the 30 year crossed 5.55 percent on September 28 and the damage spread well past the bond proxies. |
This Week in One Minute
Eight of eleven sectors fell, and five companies went up anyway.
This was a down week that looked narrow and was actually something stranger. The S&P 500 fell 0.49 percent, the Russell 2000 fell 1.02 percent, and eight of eleven sectors finished lower against three the week before. The cause was the long end of the Treasury market: the 30 year yield rose 14.2 basis points to 5.603 percent after touching 5.638 on September 30, the highest close of this cycle. What makes that reading unusual is what happened the same morning. The August inflation report came in cooler than expected, 3.4 percent against a 3.7 percent forecast, the three month bill yield fell nine basis points across the week, and traders cut the odds of an October rate increase to about 37 percent. Good inflation news, and the long bond sold off anyway. That is a supply problem, not a price problem, and it is why two Treasury auctions next week matter more than any earnings release. Against that tape, the only things that rose were companies that produced dated, specific evidence in the same five sessions: a bookings record, a raised guidance range, a court ruling, a record quarter and a funded buyback. Nothing moved on a theme this week. Everything that moved, moved on a document.
- A cooler than expected inflation print on September 30 pushed the short end down nine basis points and the 30 year up to its highest close of the cycle, which tells you the long end is being driven by supply rather than by inflation.
- All five names in this issue rose on an event with a date and a filing behind it. Fifty candidates were screened out, most of them for the opposite reason: a large move with no company announcement to attach it to.
- Three of the five entered the week as names the market had written off this year. Accenture was down about 32 percent for the calendar year and Carnival was 34 cents above its 52 week low.
The Week Ahead
WHAT I EXPECT · 66% · MODERATE
I expect the market to keep paying for a short list of companies that produced dated evidence and to keep taking it out of everything else. From the October 2 close to the October 9 close I expect the technology sector fund XLK to beat the equal weight S&P 500 fund RSP again, as it has in each of the last six weeks by 1.23, 1.62, 1.56, 2.23, 3.65 and 2.19 percentage points. The reason is the discount rate, and the reason the discount rate is rising is supply rather than inflation. On September 30 the August inflation print came in at 3.4 percent against a 3.7 percent expectation, the three month bill yield fell nine basis points across the week as the market priced October rate increase odds down to about 37 percent, and the 30 year yield rose 14.2 basis points anyway to close at 5.603 after touching 5.638, the highest close of this cycle. A long bond that will not rally on good inflation news is being sold by someone who has to sell it, and two reopening auctions land inside this window, a 10 year on October 7 and a 30 year on October 8. A rising term premium subtracts most from businesses whose cash flows are fixed and least from businesses growing fast enough to outrun it, and the average stock is the former. I put this at 66 percent rather than higher because the September employment report lands before the window opens and I am writing without it.
What would prove me wrong. The technology sector fund XLK trails the equal weight S&P 500 fund RSP by more than 1.5 percentage points from the October 2 close to the October 9 close, measured as the percentage change in XLK minus the percentage change in RSP.
Last week's call, graded. The kill condition never fired. RSP finished the window 0.05 percentage points behind SPY, and the closest it came to the trigger was 0.16 points on September 29. So the call survived. The mechanism I described did not. The cap weight lead I said would keep widening collapsed from 1.68 points to 0.18, and it collapsed for a reason that has nothing to do with my argument: three large companies in one sector fell, with Meta down 6.64 percent, Netflix down 5.40 percent and Alphabet down 1.20 percent. Breadth did not improve. The Russell 2000 fell 1.02 percent, more than either index measure, and eight of eleven sectors finished lower against three the week before. I was right about the result and wrong about the road, which is the kind of survival worth saying out loud. The risk I named is the part that deserves the attention: I said at 35 percent that the rate move would stop being selective, with the 30 year closing above 5.55 percent by October 9 as the confirming reading. It closed 5.561 on September 28, eleven days early, then 5.594 and 5.638. The risk I flagged is the thing that happened.
By how much. I said I would be wrong if the equal weight index beat the cap weight index by more than 1.0 point. Its closest approach was 0.16 points on September 29, so the call cleared its own kill line by 0.84 points, about 0.6 standard deviations. That is a genuine survival rather than a near miss, but it is not a comfortable one, and the mechanism behind it failed even though the threshold held.
And the watch list. Four of the five dated items I published last week held. The one that did not was Meta: I said it would stay above 720 dollars and it closed 715.62 on September 28. It recovered to 725.93 by the end of the window, but the condition I wrote was any close below 720, and that was reached, so it scores as a miss. Across 35 logged watch items the hit rate is 82.9 percent, and those short dated calls are the only part of this record with enough observations to mean anything yet.
What is on the calendar.
| Oct 2 | The September employment report, 8:30 a.m. Eastern. It lands the morning this issue is delivered, so nothing here is written with knowledge of it. A hot number puts the October rate increase back on the table and extends the sorting; a weak one is the fastest way the laggards get bid. |
| Oct 5 | The ISM services survey. The September 23 flash survey is what started this rate move. A second hot services reading says the September print was not a one off and keeps the pressure on the long end. |
| Oct 7 | A 10 year Treasury note reopening auction, and the minutes of the September Federal Open Market Committee meeting. Direct supply into a market that has sold off for three straight weeks, on the same day the market sees how much disagreement there was behind the September rate increase. |
| Oct 8 | A 30 year Treasury bond reopening auction. This is the one. The 30 year is the observable behind the whole week and behind the risk below. A weak auction extends everything in this issue; a strong one is what would reverse it. |
| Oct 14 | The September consumer price index. Outside this window, but it is the reading that decides whether the December rate increase the market has moved toward is real. |
THE BIGGEST RISK · 25% · BY OCTOBER 16
The auctions go well and the long end finally retreats. Every week this year that the 30 year has backed off, the names that were sold on rates have rallied hard and the concentrated winners have gone sideways. A strong 10 year on October 7 and a strong 30 year on October 8 would do more damage to what I have just written than a weak one would, because the entire expectation rests on the discount rate continuing to sort the market.
Why I think so. The 30 year yield has risen three weeks running to 5.603 percent and sits at the highest close of this cycle, which is where supply pressure is most visible and also where a buyer strike is most likely to end. The short end has already turned the other way: the three month bill yield fell nine basis points this week and October rate increase odds fell to about 37 percent from roughly a coin flip the day before. If the long end follows the short end down, utilities, real estate, financials and the Russell 2000, which have lagged for six straight weeks, are the first things bought.
What would confirm it. The 30 year Treasury yield closing below 5.45 percent on or before October 16, 2026. That is about 15 basis points below the October 1 close of 5.603 percent, a little over one standard deviation of a two week move in this observable.
Market Risk Score
64 out of 100. Elevated. Up two points from last week.
The score went up for three reasons and stayed off the highs for one. Eight of eleven sectors fell against three the week before, the average stock fell further than the index for a sixth straight week, and the long end made a new cycle high in the same week that a cooler inflation print should have relieved it. What keeps this at 64 rather than higher is equity volatility: the VIX closed at 16.39, which is not a market in distress. The stress is in the bond market and it is leaking into equities one sector at a time.
One disclosure about this number. The calibration ledger now has five matured readings of the Market Risk Score against what the market actually did next, and the relationship is backwards: higher readings have preceded stronger forward markets, with a correlation of 0.479. Five 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. It has not yet earned the right to tell you what happens next, and I would rather say that than keep printing a number that quietly implies otherwise.
Sector Rotation Update
Into: Technology · Utilities · Energy · Industrials
Out of: Communication Services · Real Estate · Materials · Health Care
| SECTOR | WEEK | TREND |
|---|
| Technology The only sector that rose meaningfully, up 1.59 percent, and a sixth consecutive week of beating the equal weight index. | +1.59% | Leading |
| Utilities Up 0.81 percent, which is a bounce off last week heavy selling rather than a trend, and the Improving label is the weakest performing label in the calibration record. | +0.81% | Improving |
| Energy Up 0.16 percent. Refining margins stayed wide but no company in the group produced a dated announcement this week. | +0.16% | Neutral |
| Industrials Down 0.11 percent, essentially flat, with the data centre linked names holding up and the transport names giving ground. | -0.11% | Neutral |
| Consumer Discretionary Down 1.37 percent, with Home Depot down 3.33 percent, even as Carnival printed a record quarter inside the same sector. | -1.37% | Slipping |
| Consumer Staples Down 1.68 percent, with McCormick falling 7.7 percent across the week despite reporting a third quarter beat. | -1.68% | Slipping |
| Financials Down 1.96 percent, with Bank of America down 4.10 percent, which is unusual company for a week that steepened the yield curve. | -1.96% | Slipping |
| Health Care Down 2.16 percent, with Johnson and Johnson down 4.44 percent, while United Therapeutics rose 21 percent inside it on a court ruling. | -2.16% | Lagging |
| Materials Down 2.29 percent, consistent with a rising discount rate taking the most from businesses whose cash flows are fixed. | -2.29% | Lagging |
| Real Estate Down 2.33 percent, the most direct casualty of the 14.2 basis point rise in the 30 year Treasury yield. | -2.33% | Lagging |
| Communication Services Down 3.55 percent and the weakest sector, almost entirely because of three companies: Meta down 6.64 percent, Netflix down 5.40 percent and Alphabet down 1.20 percent. | -3.55% | Lagging |
Is the rotation durable? While the term premium is rising, yes, and nothing on the calendar settles that before the 30 year auction on October 8. Money kept leaving the sectors whose cash flows are fixed and stayed in the one whose cash flows are growing fast enough to outrun the discount rate. Two honest caveats. The Utilities reading is a bounce off last week heavy selling, not a trend, and in the calibration record the Improving label is the least reliable one I use, right 13 percent of the time at a one month horizon across nine readings. And the rotation read overall scores 63 percent at one month across 41 matured readings with a mean edge of 0.7 points against the index, which is why I lean on the direction and not on the labels.
Breadth: What Actually Narrowed
This needs its own section this week because the obvious reading of the numbers is wrong.
The gap between the cap weight S&P 500 and the equal weight version collapsed from 1.68 percentage points last week to 0.18 this week, the sharpest narrowing in five weeks. On the face of it that says breadth improved and the market broadened out. It did not.
The gap narrowed from the top down. Three large companies inside one sector fell: Meta down 6.64 percent, Netflix down 5.40 percent and Alphabet down 1.20 percent, which carried Communication Services to down 3.55 percent and the worst sector of the week. Tesla fell 6.31 percent in Consumer Discretionary. Those are heavy index weights, so the cap weight index gave up the lead it has held for six weeks without the average stock doing anything better at all.
The proof is in the two readings that measure the average stock directly. The Russell 2000 fell 1.02 percent, more than both index measures. And the sector count went from three of eleven falling last week to eight of eleven this week. Breadth got worse, not better.
There is a second measure of concentration that did not narrow. The technology sector fund has now beaten the equal weight index for six consecutive weeks, by 1.23, 1.62, 1.56, 2.23, 3.65 and 2.19 percentage points. That is the one to watch, because it measures whether the market is paying for one kind of business rather than whether the biggest companies happen to be having a good week. The market is not broadening. It changed which small group it is paying for.
The Five
01 · RECORD BOOKINGS
Accenture (ACN) +19.67%
Technology / IT Services · Friday Five Score 86 · Fresh Breakout Accenture reported fiscal fourth quarter results on October 1 and the shares rose 15.8 percent that session, finishing the week up 19.67 percent. Fourth quarter revenue was 18.68 billion dollars and diluted earnings per share 3.29 dollars, but the number that moved the stock was bookings: 22.17 billion dollars in the quarter and 84.54 billion dollars for the full year, an all time high. For most of 2026 this stock has been sold on one argument, that artificial intelligence would shrink the work it does. The bookings figure is the first hard reading against that argument, and it arrived with the stock down about 32 percent for the calendar year.
Why it made the five. Three things separate this from the rest of the screen. The move is one session and it is the session after the results, so the cause is not in dispute. The catalyst is a forward number rather than a backward one: bookings are work signed and not yet delivered, which is the only part of a consulting income statement that says anything about next year. And the setup was a genuine disagreement, not a crowded long. A company entering its print down 32 percent for the year while the index is up 12 percent is a company where the market has taken a position, and this is the week that position got tested by data.
The case against it. The guidance does not support the enthusiasm. Fiscal 2027 revenue growth was guided to 3 to 6 percent in local currency, slower than what fiscal 2026 just delivered, and guided shareholder returns of at least 9.5 billion dollars are below the record 11.5 billion dollars returned in fiscal 2026. Bookings are also the easiest figure in this business to flatter, because a large multi year renewal lands in one quarter and says nothing about pricing. At roughly 14.5 times the midpoint of its own fiscal 2027 earnings guidance the stock is no longer priced for disaster, so the discount that made this interesting is partly gone. And one quarter does not settle a structural argument about what automation does to billable hours.
WHAT TO WATCH NEXT
Fiscal first quarter 2027 results, expected mid December though the company has not announced a date, the first quarter that tests whether the record bookings convert into the revenue growth the guidance does not yet assume
Confirms the thesis: Bookings growth holds and management lifts the 3 to 6 percent fiscal 2027 revenue range
Breaks it: Bookings decelerate, which would make this quarter a pull forward rather than a turn
Source: U.S. Securities and Exchange Commission, Accenture plc Form 8-K exhibit, fourth quarter and full year fiscal 2026 results, October 1, 2026. https://www.sec.gov/Archives/edgar/data/0001467373/000146737326000037/q4fy26earnings8-kexhibit.htm
02 · GUIDANCE RAISE
Synopsys (SNPS) +15.45%
Technology / Software · Friday Five Score 84 · Extended Synopsys held its 2026 investor day on September 30 and guided fiscal 2027 revenue to 11.1 to 11.2 billion dollars, about 15 percent growth and above the roughly 10.8 billion dollars the sell side had been carrying. It also set a long term model of about 50 percent non-GAAP operating margin by fiscal 2030, announced an intellectual property agreement with Amazon that makes Amazon the lead customer for a new application optimised intellectual property business, announced a chip design partnership with OpenAI, and said it intends to repurchase about 1 billion dollars of shares in the coming months. The shares rose 4.8 percent on the day and 12.8 percent the day after, closing the week up 15.45 percent.
Why it made the five. The design software companies sit at the front of every chip that gets built, which makes their guidance a better forward read on semiconductor capital spending than most semiconductor guidance. What makes this a pick rather than a theme is that the number was raised against a published consensus, by a company that has historically guided conservatively, and the raise came with a named lead customer rather than a market size estimate. The comparison that matters is Cadence, which rose 6.2 percent on October 1 against 12.8 percent here on the same tape. The market treated this as company specific, and so do I.
The case against it. Everything announced this week is a plan. Neither the Amazon agreement nor the OpenAI partnership carries a disclosed contract value, the fiscal 2030 margin objective is four years out, and the fiscal 2027 range is guidance the company has not yet had to defend against a reported quarter. The stock added 15 percent in two sessions for a four year model, which prices a meaningful part of that plan before any of it reaches the income statement. It now trades within about 9 percent of its 52 week high, so the margin for a disappointing fiscal fourth quarter in December is thin. And a customer concentration story that starts with one named lead customer is a concentration risk as much as a validation.
WHAT TO WATCH NEXT
Fiscal fourth quarter and full year fiscal 2026 results, expected in early December though the company has not announced a date, the first report after the new model was set
Confirms the thesis: The fiscal 2027 range is reaffirmed with the Amazon arrangement quantified
Breaks it: The company reiterates the long term targets without putting revenue behind either partnership
Source: Synopsys, Inc., Synopsys Details Growth Strategy and Long-term Financial Model at 2026 Investor Day, September 30, 2026. https://www.prnewswire.com/news-releases/synopsys-details-growth-strategy-and-long-term-financial-model-at-2026-investor-day-302894684.html
03 · PATENT RULING
United Therapeutics (UTHR) +21.00%
Health Care / Biotechnology · Friday Five Score 80 · News Spike On September 30 the United States District Court for the District of Delaware found that Liquidia Technologies infringes two claims of a United Therapeutics patent covering inhaled treprostinil for pulmonary hypertension associated with interstitial lung disease. United Therapeutics said it believes it is entitled to an order instructing the Food and Drug Administration to withdraw approval of the competing product, and that it intends to seek damages including a royalty on past sales. The shares rose 12.5 percent that session and 5.4 percent the next, finishing the week up 21.00 percent. That happened in a week when Health Care was the third weakest sector, down 2.16 percent.
Why it made the five. A patent ruling is the most literal form of proof a market can be handed, which is why it belongs in this issue. It is also the cleanest case on the list for ruling out alternatives: two other United Therapeutics items circulated in the same period, the Food and Drug Administration accepting its idiopathic pulmonary fibrosis filing and an accelerated buyback, and both were announced in early September, before this measurement window opened. The ruling is the only dated event inside it. A 21 percent move in a sector that fell 2.16 percent is about as company specific as a week gets.
The case against it. A district court opinion is not a final judgment, and the market has priced one as if it were. The remedy has not been set, whether the Food and Drug Administration is actually instructed to withdraw the competing approval has not been decided, and an appeal is available. The sentence doing most of the work in the press release is the company stating what it believes it is entitled to, which is a litigation position rather than an order. Patent outcomes in this field have reversed on appeal before. The shares now sit about 6 percent below their 52 week high, so the price already assumes the best available version of a process that is not finished.
WHAT TO WATCH NEXT
Third quarter results, expected in late October though the company has not announced a date. The one confirmed date ahead of it is April 26, 2027, the Food and Drug Administration target action date on the separate filing to extend nebulized Tyvaso into idiopathic pulmonary fibrosis
Confirms the thesis: The court issues a remedy consistent with the company stated position and Liquidia exhausts its options
Breaks it: The remedy falls short of withdrawal, or an appeal stays it
Source: United Therapeutics Corporation, United Therapeutics Corporation Prevails in Patent Litigation Against Liquidia, September 30, 2026. https://ir.unither.com/press-releases/2026/10-01-2026-000907007
04 · RECORD RESULTS
Carnival Corporation (CCL) +15.05%
Consumer Discretionary / Travel and Leisure · Friday Five Score 78 · Fresh Breakout Carnival reported third quarter results on September 29 and the shares rose 13.4 percent that session, closing the week up 15.05 percent. Revenue was 8,435 million dollars, net income 1,920 million dollars and adjusted earnings per share 1.43 dollars, all records. Net yields reached 254.99 dollars per available lower berth day, up 2.4 percent in constant currency, and customer deposits reached 7.6 billion dollars, half a billion above a prior year record. The company raised full year adjusted earnings per share guidance to 2.24 dollars, an improvement of more than 150 million dollars in adjusted net income against June guidance despite a 150 million dollar fuel headwind. The shares entered that morning at 21.79, which is 34 cents above their 52 week low.
Why it made the five. This is the issue only read on the ordinary consumer, and it is a useful one because the rest of the consumer tape disagreed with it. Consumer Discretionary fell 1.37 percent across the week, Walmart fell 3.10 percent and Home Depot fell 3.33 percent, and a cruise operator printed records anyway. Customer deposits are the reason I take it seriously: that figure is cash already paid for travel not yet taken, which makes it the closest thing in the consumer complex to a forward order book. The competing explanation, that this was a sector bounce, does not survive the sequence. Royal Caribbean rose 7.5 percent on the same day, roughly half as much, and only after Carnival opened.
The case against it. The guidance was not uniformly good. Fourth quarter adjusted earnings per share were guided to about 0.20 dollars against a consensus near 0.25 dollars, and the full year raise absorbs a 150 million dollar fuel headwind that has not gone away. The shares are still down about 18 percent for the calendar year and about 26 percent below their 52 week high, so this is a recovery from a depressed level rather than a breakout from strength, and the market has repeatedly treated record cruise results as a cycle peak rather than a trend. The business also carries real leverage into a week when the long end of the Treasury curve made a new cycle high, which is the least convenient backdrop a heavily financed operator can have.
WHAT TO WATCH NEXT
Fourth quarter and full year results, expected in late December though the company has not announced a date, covering the fiscal year that ends November 30
Confirms the thesis: Customer deposits make another record and the fourth quarter clears the 0.20 dollar guide
Breaks it: Deposits flatten, which would be the first sign the forward book is turning
Source: U.S. Securities and Exchange Commission, Carnival Corporation Form 8-K, third quarter 2026 earnings release, September 29, 2026. https://www.sec.gov/Archives/edgar/data/0000815097/000081509726000104/a20263qearningsrelease8-k.htm
05 · CAPITAL RETURNS
Talen Energy (TLN) +7.69%
Utilities / Independent Power · Friday Five Score 74 · Watch Pullback On September 29 Talen Energy named Terry Nutt chief executive officer effective January 1, 2027, and announced in the same release that it had entered 1.5 billion dollars of accelerated share repurchase agreements that day and upsized its total repurchase authorisation to 3.0 billion dollars through 2028. The company said the program is expected to retire more than 10 percent of its shares at current prices. It is funded in part by about 1.5 billion dollars of PJM capacity revenue monetisation for the 2027 to 2028 and 2028 to 2029 delivery years, completed four days earlier on September 25. The shares rose 7.69 percent across the week while the Utilities sector rose 0.81 percent.
Why it made the five. Almost everything in the power and data centre complex this year has been a promise about demand in 2028. This is the opposite: the company sold future capacity revenue on September 25 and spent 1.5 billion dollars of it buying its own stock four days later. Cash has moved and the share count is going down. In a week when the market was paying for evidence and ignoring narrative, that distinction is the whole reason this name is here rather than one of the larger independent power producers that moved on headlines.
The case against it. Monetising capacity revenue for the 2027 to 2028 and 2028 to 2029 delivery years converts future cash into present share count reduction. It raises per share figures without adding a megawatt of generation, and it leaves less cushion if power prices or PJM capacity prices move against the company in exactly those years. A chief executive transition announced alongside an active capital allocation program is its own execution risk, since the person who set the plan is not the person who will finish it. The stock is still down about 14 percent for the calendar year and about 28 percent below its 52 week high, and the 7.69 percent move is roughly the size of the share count reduction, which suggests the market took the arithmetic at face value and nothing beyond it. This is also the thinnest name in the issue at about 277 million dollars of average daily trading.
WHAT TO WATCH NEXT
Third quarter results, expected in early November though the company has not announced a date. The confirmed date is January 1, 2027, when the chief executive transition takes effect
Confirms the thesis: The share count falls visibly in the third quarter filing and the repurchase program continues at pace
Breaks it: The repurchase slows, or the capacity monetisation proceeds are redirected
Source: U.S. Securities and Exchange Commission, Talen Energy Corporation Form 8-K exhibit, chief executive appointment and capital return, September 29, 2026. https://www.sec.gov/Archives/edgar/data/0001622536/000162253626000074/a20260929pressreleasecapac.htm
Stock of the Week
Accenture (ACN)
Accenture is the Stock of the Week because it is the clearest example of what this week was about. Going into October 1 the stock was down about 32 percent for the calendar year while the S&P 500 was up about 12 percent, a gap of more than 40 points, and the reason was a single argument: that artificial intelligence would eat the demand for technology consulting. That argument had never been tested against a number. On October 1 it was, and the number was 84.54 billion dollars of full year bookings, an all time high. The stock rose 15.8 percent in one session. The lesson is not that the bear argument was wrong, because one quarter does not settle a structural question, and the company own fiscal 2027 revenue guidance of 3 to 6 percent growth is slower than what it just reported. The lesson is about what a long, unresolved disagreement does to a price. When a stock has been sold for nine months on a thesis that nobody has been able to check, the first checkable piece of evidence moves it violently, in whichever direction it points.
What Could Change My Mind?
A pair of strong Treasury auctions. This is the one. If the 10 year on October 7 and the 30 year on October 8 both clear with ordinary demand and the 30 year yield slips back under 5.45 percent, the whole argument in this issue reverses within days. The names that have lagged for six weeks lagged because of the discount rate, not because of their own results, so they are the first things bought when it eases. I put that at 25 percent.
A soft September employment report. It lands the morning this issue reaches you, so I am writing blind to it. A weak number does two things at once: it pulls the December rate increase further out and it raises the question of whether the economy behind these record bookings and record cruise quarters is still intact. A genuinely bad print would make the forward call wrong for the right reason.
Bookings that do not convert. Two of the five names in this issue, Accenture and Synopsys, rose on forward numbers rather than delivered revenue. A bookings record is work signed and not yet done, and a fiscal 2027 range is a plan. Both report in December. If corporate technology budgets soften in the fourth quarter, both give it back together, and they are the first and third largest movers on this list.
A remedy that falls short. United Therapeutics moved 21 percent on a district court finding, not on a final judgment. If the remedy is narrower than the company expects, or an appeal stays it, the most literal piece of proof in this issue turns out to have been priced for an outcome that has not been granted.
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 September 24 close to the October 1 close, against the S&P 500 fund at minus 0.42 percent:
| PICK | LABEL I GAVE IT | SCORE | WEEK AFTER |
|---|
| VLO Valero Energy | Avoid for Now | 76 | +6.69% |
| OKTA Okta | Extended | 79 | +2.90% |
| NEE NextEra Energy | Avoid for Now | 68 | +0.97% |
| ILMN Illumina | Extended | 71 | -2.99% |
| META Meta Platforms | Extended | 84 | -6.64% |
Three of the five beat the index and the group averaged plus 0.19 percent against minus 0.42 percent for the S&P 500 fund, so the week was fine. Read the two middle columns together, though, because they are embarrassing. The two names I labelled Avoid for Now were the two best performers. The name I scored highest, at 84, was the worst performer by six points.
That is not a fluke, and the ledger now says so. Across 25 matured picks the rank correlation between the Friday Five Score and the forward one month return against the index is minus 0.005. Not weakly positive. Zero, with a negative sign on it. Picks scored 85 and above have averaged plus 4.82 percent against the index at one month, and picks scored below 75 have averaged plus 7.56 percent. The low scores have beaten the high scores. I am going to keep publishing the score, because hiding a number that is not working would be worse, but until that relationship turns I would read the catalyst and the bear case and treat the score as the least informative thing on the card.
The full record, two ways. Measured in the week each pick appeared, 70 picks across 14 issues have returned an average of 12.76 percent against 0.21 percent for the benchmark, with an 85.7 percent win rate. 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, not of the idea. The honest number is the one that starts the day you could have acted. Measured from publication forward one week, 65 picks across 13 issues have returned an average of 0.38 percent against 0.44 percent for the benchmark. That is minus 0.06 points relative, a 60 percent win rate and a 52.3 percent beat rate. Over thirteen weeks the forward record is a coin flip. The headline number is marketing; the forward number is the truth, and I am going to keep printing both side by side until one of them changes.
September Monthly Leaders. This is the first issue of October, so the monthly board rolls. These are the five best performing names covered in any September issue, measured from the August 31 close to the September 30 close, in a month when the S&P 500 fund fell 0.58 percent and the equal weight index fell 5.18 percent:
| NAME | CATALYST | SEPTEMBER |
|---|
| NTRA Natera | Clinical Data | +28.37% |
| ILMN Illumina | Index Inclusion | +28.11% |
| META Meta Platforms | Adoption Data | +26.70% |
| SIG Signet Jewelers | Guidance Raise | +24.37% |
| OKTA Okta | Standards Coalition | +20.84% |
Two things are worth saying about that board. The first is that four of those five catalysts were the same kind of thing this issue is about: a dataset, an index decision, a download ranking, a guidance raise. Each one was a specific, dated, checkable event. The second is that September was a month when the average stock fell 5.18 percent, so this board is not a reflection of a rising market. It is also, honestly, a selection of the five best out of twenty names covered, and the other fifteen are on the Scoreboard page where you can see all of them.
The screens, graded. I reject far more names than I publish, and those rejections are tracked now too. Across 201 logged rejections, 71 percent of the names I screened out went on to underperform at one month: 45 correctly avoided, 20 neutral, and 18 that were real winners I passed on. The single most reliable screen has been the catalyst check, where 100 percent of the nine names rejected for having no verifiable catalyst went on to lag. That is the screen doing most of the work in this issue, where 50 of 55 candidates were rejected and the largest group of them were names that moved a long way on nothing I could point to.
Terms in This Issue
Accelerated share repurchase. A deal in which a company pays a bank a large sum up front and immediately receives most of the shares it is buying back, instead of buying them gradually in the open market over months. The bank sorts out the actual purchasing afterwards.
New bookings. The value of work a services company has signed but not yet performed. It is a forward indicator, because revenue in later quarters has to come out of it, though the timing and the profitability are not fixed when it is signed.
Term premium. The extra yield investors demand for lending money for thirty years instead of rolling over short loans. When it rises while inflation expectations fall, the bond market is worrying about supply and risk rather than about prices.
Bear steepener. A move in which long term interest rates rise faster than short term rates, so the gap between them widens while both ends are under pressure. It usually signals concern about government borrowing rather than about central bank policy.
Equal weight index. A version of an index that gives every company the same weighting, so the average company drives the result rather than the largest few. Comparing it with the standard index is the quickest way to see whether a market move is broad or narrow.
Net yield. For a cruise operator, the revenue earned per available bed per day after the costs directly tied to carrying that passenger. It separates genuine pricing power from simply sailing more ships.
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.
|