On October 19th these pages argued that the base rate was skeptical the July 29th low would hold, and that the signal for a real bottom has been credit spreads peaking.

The bull case was that the stress sat with the leveraged builders while the hyperscalers funded themselves.

Meta’s Q3 report on October 21st undercut that bull case.

• SMH fell 6.0% the next day to $523.35, ‑21.8% from its June 22nd peak.

• The July low held on the equity side but AI credit blew out and the market priced immediate dilution.

Then, today, Monday morning, Meta launched ~$35bn of new common and mandatory convertible preferred stock, with a ~$60bn at-the-market program to follow. The total ask: as much as $95bn.

Bottom line up front:

• The funding leg broke at the top. Meta reported negative free cash flow, more debt than cash, and a new share class “usable for capital raising.” The self-funded tier now needs the capital markets. On Monday it went to them for up to $95bn, with Nvidia, AMD and Broadcom lined up to buy.

• AI credit marked the July low. On July 29th, Meta, Oracle and CoreWeave CDS all peaked the same day SMH bottomed. None of them bought the 25.7% equity bounce rally, and the referenced credits all are at or near their widest now.

• The July low held. SMH’s October 22nd low held 3.8% above July’s, sitting just above the nearest downside scenario level ($516).

• A bounce is likely and should be treated as such. Forced selling, washed-out semis breadth and late-October seasonality favor a bounce into the October 28th–29th prints. Until AI credit turns, it’s just a bounce.

• The position. I held the SMH long through Meta’s print. Why the size was wrong is in Self-Awareness is a Superpower.

• This note is about what comes next.

What Meta said

Meta’s Q3 was a strong quarter with one problem: the cash didn’t cover the build.

• Revenue $64.6bn, up 26%. Capex $34.9bn. Free cash flow negative $2.1bn.

• $83.7bn of debt against $82.4bn of cash and securities. A year ago it was $28.8bn of debt against $44.4bn of cash.

• 2026 capex $135–145bn, top end unchanged. For 2027, “continue investing aggressively,” with numbers on the Q4 call.

• A new non-voting share class, trading from late January 2027, “usable for capital raising, employee compensation, acquisitions.” The CFO: “We don’t have any specific financing plans to announce at this time.”

• META fell 7.0% on October 22nd and 9.4% on the week. SMH fell 6.0% that day.

What Meta did Monday

Read more

October has been my worst month of the year: down about 9%, most of it from five trades I put on September 21. I wanted to know why, so I went back through everything.

Over the past few weeks I re-ran every trade in my Ledger since August 2024, about 270 of them, against the same markets on random dates. I also tested 35 markets over more than 20 years t…

Read more

On August 3rd I looked at what history says about SMH after a ‑20% correction. SMH had just crossed ‑20% (July 28th) and bottomed at ‑24.6% the next day.

The conclusion was that the base rate argued for skepticism that the July 29th low would hold.

Since then SMH rallied 25.7%, to within 5.2% of its June 22nd high on October 5th, then gave almost all of it back in the AI credit scare covered in yesterday’s note. It fell 10.5% last week, its fifth-worst week since 2011, on nine straight down days, the longest losing streak in its history. Friday’s close was $539.79, ‑19.3% from peak.

Bottom line up front:

• History still says lower. Every completed ‑20% episode made its final low later and lower than the crossing. The two that rallied back to ‑5% (2015-16 and 2024-25) went on to new lows.

• Credit is the tell. Every SMH final low since 2011 came within two trading days of the peak in single-B or CCC spreads. Both made new highs for this episode on Friday.

• The bull case is that credit peaks early. The stress this time sits with the leveraged builders while the hyperscalers fund themselves. If that holds, spreads can peak well below historical levels and the low is near.

• No policy rescue. Three of the five prior lows came with a policy backstop. A cut, or a signal that the hiking cycle is one and done, looks unlikely with inflation currently where it is. That leaves the hyperscalers’ 2027 capex guides on October 28th and 29th as the catalyst.

• Staying long SMH into the prints. The capex guide decides which scenario plays out, semis carry the strongest estimate revisions into it, and SMH and MAGS hedge each other on the guide. The scenario table below sets out the downside if it goes wrong: $436 to $378.

Forward returns from the ‑20% crossing

For each ‑20% SMH correction since inception (December 2011), I find the trading day the drawdown first crossed ‑20%, then measure returns 5, 10 and 21 trading days later, plus how much further SMH fell to its final low and how long that took.

Averages across the 5 completed episodes:

• 1wk mean ‑0.6% (median +2.9%, positive 60% of the time); 2wk mean +5.0% (median +10.0%, positive 80%); 30d mean +3.7% (median +3.1%, positive 100%).

• Further decline to the final low averaged ‑14.8% (range ‑2.5% to ‑28.6%), over 99 trading days on average (range 9 to 170).

• 2026 fit the short-term pattern: +8.7% after one week, +8.2% after two and +4.9% at 30 days.

30-day paths indexed to 100 at the ‑20% crossing. Larger versions are at the end of the note.

The failed rally

Read more

The two notes that preceded this one, the October 7^th Signal & Noise Filter and the October 13^th Market Commentary, pointed towards cyclical bullishness balanced on a narrow set of conditions, mainly that the AI secular trade stays on track.

Last week the S&P 500 fell 2.0%, the Nasdaq 4.5%, and SMH 10.5%, the worst week for chips since April 2025.

Thus, “we are so back” has given way to “it is so over”.

I. What Happened

• Korean memory broke first. SK Hynix fell 10.1% and Samsung 7.4% on October 13th, Korean time, before the US open and before CoreWeave. By Friday, Hynix was down 28% since September 30th and 57% from its June 22nd peak.

• Then AI credit cracked. On October 14th CoreWeave’s 9⅝% 2032 bonds hit 77.8, down from 100.9 in June, and the stock fell 16%. A Gartner estimate that data-center owners need $8.5 trillion of revenue through 2030 against $3.6 trillion forecast landed the same day. Oracle’s 5-year CDS ended the week at 300bp, within about 30bp of the whole high-yield index, for a BBB- hyperscaler.

• Inflation came in hot, and yields fell anyway. Core CPI rose 0.3% and core PPI 0.5%. The 10-year still fell to 5.09%. The bond market traded growth, not inflation.

• Oil kept going. December Brent made a new high at $109.86 on Friday and WTI closed the week back above $100, the same week a cargo ship hit a mine and sank in the Strait of Hormuz.

Figure 1. Bloomberg (000660 KS, 005930 KS, 2330 TT, local currency, local dates) and Pinebrook dashboard (SMH, SPY), Sep 30 = 100, through Oct 16, 2026.

II. What the Roadmap Said

• The AI Test Is about Growth Rates and whether AI capex spending keeps accelerating. An AI capex wobble was flagged as problematic.

• Financing was on the table. The October 13^th note’s bull/bear table listed financing as one of the AI trade’s three pillars. Thus, OpenAI’s run-rate near $50 billion against $70 billion signaled, and AI debt issuance down from $113 billion in June to $23 billion in September are now flashing yellow lights.

• The growth-scare row described most of last week. The October 7th outcome table: health care and long Treasuries lead; semis and small caps lag. Not in the bingo card was energy rallying instead of lagging.

• The rates reaction was called in real time. In the subscriber chat before the AI numbers came out: if the one thing driving the economy sputters, 10-year yields fall. They did, through two hot inflation prints.

• The inflation call held. The preview had core CPI at .26% against a .2% consensus; it printed .3%. Core PCE is now tracking .34% against a street .32%.

• The XLE short was covered at Tuesday’s close. Brent has made a new high since.

III. What I Got Wrong

• The positioning table mapped the wrong shock. It ran CPI outcomes through rates and sectors. Instead, the shock came through AI credit. Hot inflation with falling yields was not an expectation that made the short list.

• Junk was read as a rates story. The October 7th note said CCC widening was a discount-rate shock, not a growth warning. CCC has widened another 84bp since, and single-B has followed it up.

IV. What Is Driving It

Read more

Subscribers: open the dashboard at dashboard.pinebrookcap.com →

Today’s dashboard upgrade starts with security, and it also puts the dashboard on your phone.

A more secure sign-in

The dashboard now signs you in with a magic link, and each sign-in lasts 30 days.

1. Go to the dashboard and enter your email as usual.

2. Check your inbox for an email from Pinebr…

Read more

Executive Summary

• September core CPI (Oct 14) printed .33% MoM (.335% unrounded) vs. a .2% consensus and Pinebrook’s .26% call. Core YoY 2.6%.

• Headline CPI rose .65% unrounded on a 9.1% jump in gasoline, taking YoY to 3.7%. The Pinebrook gas nowcast (.61%) was 4bps light, well inside its 20bp average miss.

• The miss was broad, not shelter: supercore +.42%, core goods +.25% (largest in a year). Widespread but moderate price pressure.

Read more

Sunday’s inflation preview was the easy part. The hard part is navigating the next three weeks, which include:

• Wednesday & Thursday, October 14 and 15, CPI and PPI, respectively, and thus core PCE.

• Wednesday, October 28: FOMC decision. After the close, Microsoft, Alphabet and Meta.

• Thursday, October 29: Amazon and Intel, with Apple that day or the following Monday.

• Friday, October 30: the ECI, which tests the wage leg of the inflation story.

Bottom line up front:

• Pinebrook’s CPI call is above consensus. Position for the chop to continue after the CPI print.

Read more

• Headline CPI is projected at .61% MoM versus a .6% consensus (3.6% YoY).

• See the gas-derived CPI nowcast here on the Pinebrook Dashboard.

• September core CPI (Wed, Oct 14) is projected at .26% MoM versus a Bloomberg consensus of .2% MoM and 2.5% YoY.

Read more

• Pinebrook generated a +18.60% return for the period January 2 through September 30, 2026, against an S&P 500 return of +11.77%, for a spread of +6.82%.

• September produced a -2.45% return against the S&P 500’s -0.45%, a -2.00 percentage-point spread after August’s +7.79% month.

• The September setback followed a meaningful book reset: the Japan pair, rates exposure and gold-miner positions were the principal headwinds, while TSM, IGV and the industrial short basket provided substantial offsets. Ending NAV was $1.982 million, still +18.60% YTD and +98.21% since inception on August 5, 2026.

EQUITY CURVE VS. S&P 500

PERFORMANCE SNAPSHOT

RISK METRICS — YTD THROUGH SEPTEMBER 30, 2026

MONTHLY RETURN BREAKDOWN

MONTHLY RISK METRICS — 2026 (PART I)

MONTHLY RISK METRICS — 2026 (PART II)

Volatility, drawdown, win rate, capture ratios, best and worst observations by month. YTD row is calculated from the nine monthly return observations through September.

PORTFOLIO COMMENTARY

January – August: Summary

The portfolio entered September after a +7.79% August, its strongest risk-adjusted month of the year. The July 13–16 repositioning had materially improved the book, but the September result shows that the portfolio remained exposed to several concentrated factor moves, particularly gold miners, Japan and rates.

September: -2.45%

September produced a -2.45% return against the S&P 500’s -0.45%, a -200 basis-point spread. The month was volatile: the reconstructed daily series produced a -4.10% maximum drawdown, with the worst day on September 28 (-1.99%) and the best day on September 21 (+1.60%).

The clearest positive trade-level contributors were TSM (+$33,963) and IGV (+$25,853), both closed September 22. The GEV and CAT shorts also generated +$7,594 and +$11,177 respectively. Those gains were partly offset by the Japan pair: the DXJ longs generated about +$16,141 while the EWJ shorts lost about -$29,300, for a net Japan-pair drag of roughly -$13,159.

Gold miners also reversed part of August’s gains. The four GDX/GDXJ tranches lost roughly $28,525 on the September move, while MAGS gained about $12,219, leaving the existing gold/mega-cap basket down approximately $16,306. Rates were another material headwind: the two ZN positions and the ZT position produced combined September losses of approximately $25,540.

The book was repositioned again on September 21, adding SMH and XLK longs, an XLE short, an XLV short and an NQ futures long. Those new positions were collectively up approximately $12,311 by month-end. The September result therefore reads less as a wholesale thesis failure than as a month in which several previously profitable concentrations reversed at the same time.

ON THE RISK METRICS

September’s monthly metrics are negative across the board, with a Sharpe of ‑1.94 and Sortino of ‑2.87. Beta of .95 and correlation of 0.70 show the book moved closely with the market day to day, but the S&P 500 fell only ‑0.45% for the month; annualized alpha of ‑24.1% shows that most of the loss came from position-specific moves in rates, gold miners and Japan rather than from market direction.

YTD metrics softened from their restated August 31st levels: Sharpe from 1.20 to 1.17, Sortino from 1.18 to 1.09, annualized alpha from +14.8% to +12%, and downside capture from 108.6% to 116.4%. The maximum drawdown of ‑10.37% (July 23rd) is unchanged; September’s ‑4.1% intra-month decline did not approach it. Beta moved from 1.08 to 1.14 and correlation from 0.58 to 0.63. The portfolio ended September ‑2.45% below its August 31st month-end high.

TRACK RECORD

Portfolio performance since inception in August 2024.

Everyone is asking whether a Santa rally shows up this year, or if the market continues its sideways chop as the multiple continues to bleed out and reveals a lump of coal under the tree.

That is the noise.

The signal is found by exploring the conditions that bring about one or the other.

Read more

They answered. Then I told them what I heard, and what I’d build.

It’s done. And it’s better than I expected.

Here’s what they asked for, and what they’re getting.

First, the X problem.

• The private X feed is now dormant.

• Discord replaced it: gated, paid-only, access granted when you subscribe and revoked when you don’t. The macro reads and real-time commentary live there now.

Second, the blotter.

• Subscribers wanted a trade record that’s easy to read.

• When I enter or exit a position, it goes out in Discord and the Substack private group chat.

• And the Ledger is live: every position, timestamped, measured against the S&P, with the equity curve and drawdown on the same page.

• No screenshots. No hunting.

• And it updates automatically.

Third, the dashboard.

• Not a low-rent scaled-down fake Bloomberg Terminal. Not another dressed up Yahoo Finance feed.

• A systematized, always-current view of where things stand, built on the tools I actually use.

That’s what this is: my entire framework, transparent and automated.

• Rates. Every move in the 10-year split into real rates, term premium and inflation, and what drove it. How the whole curve moved. How much Fed tightening or easing the market is pricing. Whether policy is restrictive or accommodative relative to neutral.

• The Fed, beneath the headline. The method behind my November 2023 soft-landing call, automated and backtested. Since 2015, when the Fed quietly stopped projecting a recessionary labor market, the 2-year fell an average 47bp over the next three months.

• The economy. A weekly scorecard of where we are in the cycle. The ISM regime. The labor market. And the housing-cycle work I’ve written about for years: is housing dragging on growth, is anything offsetting it, and is credit worried?

• Equities and metals. Correlation regimes, leadership breadth, the equity risk premium, and where silver and gold sit against their own history.

Every number updates automatically. Rates refresh at the Treasury close, every day. Every panel tells you how it’s built and what it can’t tell you. No black boxes. A user guide on the home page maps it all to the five moves of the framework.

When I first asked, I floated the dashboard as part of a premium tier. I built it into the core subscription instead, at no additional cost to the people who were here first.

When I built this, I kept asking one question: does this feed a signal I’d act on? If the answer was no, it didn’t make the cut.

What’s left is the framework itself, running every day where you can see it.

To my paid subscribers: you got what I promised. You came first, and you stay first. Your price is locked for the life of your subscription, whatever happens to it from here. And there’s more coming for you, in Core and in Pro, which is next in the pipeline.

To everyone reading for free: the offering just changed, and the price will change to match it. At 12:01 a.m. ET on Monday, October 5, the subscription price goes up for new subscribers.

Join before then and you lock in today’s price for as long as you stay.

Thank you all. It’s been a helluva journey.

-David

PS. Folks who are undecided can get a glimpse of what’s in the dashboard by reading a few excerpts from the dashboard user guide here.

Subscribe now

Word has it that yesterday’s move in 10-year rates was like an earthquake that shook financial markets.

• Real yields up 13bps (2.63% to 2.76%).

• Nominal yields up 15bps (4.96% to 5.11%).

Sigma-scoring a single day’s move depends heavily on which lookback window supplies the volatility estimate, so the table below runs it four ways.

10Y Daily Move Distribution, By Lookback Window

• The pre-2020 window (lowest volatility of the four) scores yesterday’s move as the most extreme, 2.7–2.9σ.

• The post-2020 window (highest volatility, shaped by COVID and the 2022 hiking cycle) scores it the mildest, 2.3–2.4σ.

• The full-history and 2023-onward windows land exactly together on the real-yield read (2.61σ both), despite drawing on different sample compositions.

Shock Aftermath: Do These Moves Mean-Revert?

A sigma score says how unusual a move was. It says nothing about what happens next.

• To answer that, every shock day in the full 2009 - present sample (|move| ≥ 13bp real, ≥ 15bp nominal, the same threshold used above) was isolated, split by direction, and tracked forward 1 to 60 trading days from the shock-day close.

• Each result was also run through a one-sample t-test against zero, because most of what falls out of an exercise like this is noise.

Signal

• Downward real-yield shocks mean-revert, and it’s statistically real. A sharp drop in real 10Y yields (average -18.9bp) has historically been followed by an average +15.4bp recovery over the next 60 trading days, basically retracing most of the move. Tested against zero: t=2.25, p=0.03. Tested against the more rigorous benchmark (the unconditional drift in real yields over any 60-day window in this sample, which is close to flat (+0.85bp)) the excess narrows slightly to +14.6bp but stays significant: t=2.12, p=0.04. Two-thirds of these episodes showed at least partial reversion by day 60.

• The reversion is a quarter-length process. At 1 and 20 days out, the average path is still negative or flat and the recovery is concentrated in the 40 - 60 day window, not the days immediately following the shock.

• The asymmetry is the other takeaway. Down-shocks in real yields behave differently from up-shocks in the same series. That directional split is a signal.

Noise

• Real 10Y up-shocks look like continuation and not reversion, but it isn’t statistically proven. Average cumulative change kept building in the same direction out to 60 days (+8.6bp further), and only about half of episodes showed any reversion at all. Directionally consistent but p=0.26 means the sample doesn’t prove it.

• Nominal 10Y shocks show no reliable pattern in either direction. A small 1-day fade after up-shocks fully unwinds by day 20 and the average just wanders from there, essentially flat by day 60, p=0.98. Down-shocks are equally directionless (p=0.41). Nominal moves include breakeven/inflation-expectation noise on top of the real-rate signal, which likely explains why they’re harder to read and noisier than the real series.

• Sample sizes are thin enough that a single episode can move the average. Each bucket is 43 - 50 observations spanning very different macro regimes: the 2009 crisis, the taper tantrum, COVID, this hiking cycle.

Applied to yesterday

• Yesterday was an up-shock in both series.

• The one statistically robust pattern in this whole exercise (real-yield down-shock reversion) doesn’t apply here.

• The directionally-suggestive-but-unproven pattern that does apply (real up-shocks tend to extend rather than revert) argues for yields staying elevated or drifting further over the next month or two, but the data doesn’t support a confident call.

Where Inflation Fits In

Read more

Introduction

Last week’s FOMC policy rate decision and Fed Chair Warsh’s subsequent press conference are now in the rear-view mirror. After a bit of flirting with near-term volatility, the market has moved on. So should we, after a parsing of the receipts.

The pre-summer thesis that led up to the above was laid out back in May, in Cruel Summer, with a projected increase in U.S. 10-year nominal yields because of:

• Broad price pressures that were moving up. ✅

• A reacceleration of the labor market. ✅

• Labor market price pressures found in healthcare benefits, not wages. ✅

• Rising earnings projections that were symptomatic of an expanding economy. ✅

• The sequencing of the Fed’s policy stance from pause to neutral to tightening over the course of the summer, validated by Chair Warsh in Jackson Hole in late August. ✅

These ideas were further examined a few weeks later in The Calm Before on May 30, where August was highlighted as a window of vulnerability and of maximum stress in the rates market.

Off by two weeks, but close enough is still a win.

While the semi trade was flagged as stretched at 4–5 sigma, the L needs to be taken on not having flagged that as a sell signal. It was indeed a sell, and a big miss here. ❌

• The semi unwind was called a rotation on June 8th: money moving between sectors rather than what it actually was, which the start of a multiple compression.

• The bond market broke bad, but did so by an order of magnitude greater than what these pages expected.

• The Cruel Summer framework and its follow-ups correctly called the Fed’s path from pause to neutral to tightening, along with rising earnings expectations.

• What it missed was drawing out the implication of its own rate call on equities.

• The multiple did exactly what the rate path predicted it would do.

While it’s early days, the data footprint is pointing toward an avoidance of the inflationary regime shift, the real Baba Yaga these pages have spent the summer watching for.

Now that the world’s most important FOMC meeting has come and gone without the world ending, it’s time to take stock of where we are in the cycle as the market front-runs the pricing of risk assets into 2027.

I. The Tape Today

• SMH and MTUM are fully normalized, at 0.03σ and 0.51σ, respectively.

• Whatever was left of the semis overextension from summer is gone.

• XLK cooled from 2.14σ to 1.55σ but didn’t fully reset.

• IGV: software is now the most stretched name of the four, peaking near 2σ on the FOMC day itself and still elevated at 1.59σ.

• Hardware/semis have normalized while software has gotten more extended over the same window.

A sectoral anomaly was flagged in the June 8 note: healthcare. By July 31st, healthcare’s correlation to the index had fallen to -0.389, the most negative reading in its entire history back to 1999.

What that divergence represented was a crowded trade with money moving into healthcare as a defensive, rate-sensitive parking spot during the tech unwind and extreme enough to push correlation to its lowest reading on record.

• A trade stretched that far eventually runs out of new money, and that’s what started happening the moment the correlation bottomed.

• From the July 31st trough, XLV’s correlation climbed in a long, almost uninterrupted grind — to -0.21 by mid-August, to zero on September 8th, and to +0.237 by September 18th.

• Healthcare took the entire round trip alone and has now rejoined the group it broke away from.

• A single sector trading with a correlation this negative to its own index is a crowded trade that usually unwinds.

Based on the evidence above, the hardware-and-momentum leg of the summer rotation is over. A narrower one in software looks like it may just be getting started.

II. The AI-Capex Path into 2027 & Beyond

Read more

I. Why This is Not a Forecast

Historical rate-cycle comparisons work the way cloud-shapes work.

• Stare at a cloud long enough and it will eventually look like a rabbit or a unicorn.

• Stare at a rate hiking cycle long enough and you’ll find one that looks like today’s, or the one your imagination prefers.

That doesn’t mean today’s cycle is 2004, or 2015, or 2022. It means a pattern-seeking mind found a shape it wanted to find.

Lean on that shape as if it were a forecast, and the market rinses you the moment this cycle turns out different, which it will because every cycle differs from every other one.

That’s the discipline of this note: historical look back windows are context, not signal.

• They answer a narrow question: is today’s number large or small relative to some comparable past, and has a particular factor shown up before? And nothing more.

• They cannot tell you what happens after Wednesday’s decision, because no two Fed committees, no two starting curves, no two sets of political and fiscal conditions are the same.

What decides the outcome is today’s data and today’s FOMC’s actual reaction function, not what a different committee did with a different curve in a different economy.

Thus, this note is not a forecast. It is a framework for outcomes under genuine uncertainty. Where the historical record is used, it is used to check whether something is stretched or ordinary, not to imply a path forward.

Therefore, tomorrow’s decision is discussed as a branching set of possibilities with a trigger and a risk for each.

II. Defining the Look Back Window

It’s tempting to look at every single Fed hike cycle back to whenever.

That will not be done to you, dear reader.

Why Not Pre-2000

The 1986-87, 1988-89, and 1994-95 cycles predate inflation targeting, predate the modern Summary of Economic Projections process, and predate a Fed reaction function shaped by two decades of disinflation, the 2008 crisis, and a zero-rate era.

The global monetary system those cycles traded in - pre-euro, pre-China-as-a-reserve-manager, a structurally different Treasury market - doesn’t match today’s plumbing.

Including them doesn’t add data as much as it adds noise dressed up as sample size.

The Sample

• Primary sample (n=3): 2004–06, 2015–18, 2022–23. Every multi-hike cycle since the Fed’s inflation-targeting era matured and the modern SEP process existed in something like its current form.

• Robustness cut (n=2): 2015–18 and 2022–23 only, because both started from the zero lower bound with a QE overhang.

Of the three, 2004–06 is the closest structural analog to today. It’s the one cycle that started from a normal, non-crisis, non-zero-bound base, with a steep positive curve, closest to the actual starting conditions in front of the Fed tomorrow.

2022–23 started from a totally different place (deeply negative real rates, a curve about to invert) and 2015–18 started at the zero bound.

III. What the Record Shows on the Narrower Window

Figure 1. SPY total return following the first hike of each cycle, post-2000 sample only (n=3).

• 4-week mean: -3.1%.

• Only two of three cycles were still positive at 26 weeks. The 26-week mean drops to essentially flat (-0.3%) once 2022’s -13% is weighed against 2004’s +8% and 2015’s +4%.

• The 52-week mean holds modestly positive (+4.0%, 2 of 3).

Financials and Utilities

Across the sample size, financial and utilities they were the worst and best-performing sectors after a first hike. And it’s not even close. They are literally the bottom and top of the list every time.

Financials negative in all three cycles at 4 weeks. Utilities positive in all three.

Rates: The Selloff Front-Loads, and This Curve Doesn’t Match Any Precedent

By the time a hike actually lands, the curve has usually already done its front running pricing work. The hike itself is a confirmation exercise, not news, unless the Fed is playing catch-up as in 2022.

Thus, 7–10Y Treasuries sold off into the first hike in all three cycles, then rallied after it in two of three, 2022 being the exception.

Now that said, today’s curve doesn’t match the shape behind any of the three historical samples. Pulling the actual pre-hike curve for each of the three cycles (6/30/04, 12/16/15, 3/16/22) against today’s gives the genuine apples-to-apples curve comparison.

Today’s curve sits above every prior hiking-cycle onset, at every maturity.

• The gap is starkest at the front end: 3-month bills sat at 1.33%, 0.27%, and 0.44% on the eve of the last three cycles; today they’re at 4.11%.

• The Fed is about to hike from an already-elevated policy rate, not from near-zero or modest levels like every comparable cycle onset.

• 2004’s 30-year point doesn’t exist, as Treasury discontinued that series between February 2002 and February 2006.

IV. The Regime Question

Last week’s note already built the case that the real question isn’t the yield level but whether the 25-year low-beta regime is ending.

That hasn’t changed. What’s new since then is a way to put an actual number on the gap.

Read more

Introduction

In the namesake film, John Wick is nicknamed “Baba Yaga”, which the film translates to “boogeyman” from Russian.

The film got the translation wrong.

• In Russian folklore, Baba Yaga is not the boogeyman. Baba Yaga is in fact an eccentric forest witch who lives in a hut.

• The real mythical night thief of bad children is Babay, which loosely sounds like baby in English and is not as frightening of a depiction of our professional assassin hero.

• These pages will take artistic liberty and correct the film’s linguistic mistake by sub-titling this note Бугимен, which is the Russian word for boogeyman (bugimen).

The inflation picture emerging from this week’s release of PPI and CPI is not the financial market boogeyman.

The financial boogeyman is whether we are looking into the abyss of a regime shift or simple reversion to the mean in rates markets, and by extension, a continuation of the rally in risk assets despite the summer stall.

Executive Summary

• Rates: Nominal 10’s are 2bps shy of their October 2023 peak, while real 10’s crossed that line on September 10th.

• The real question isn’t the level, it’s the regime. A cross-history look at 10-year yields versus nominal GDP growth shows three regimes. The current reading is uninformative and useless because it’s consistent with either a continuation of the last 25 years’ anchored regime or the early stage of a shift toward a higher one. A FOMC voter has framed the Fed’s next decision as a bet between those same two worlds.

• The hike is no longer the question, with the path and destination becoming the central risks. Since the Fed-Funds rate became the Fed’s primary tool in the 1990s, there has been exactly only one one-and-done hiking cycle. Thus, markets repriced from roughly two cumulative hikes through next June to at least three this week alone. The open questions are how many, whether the Fed needs to actively lean against nominal GDP, and whether hikes even reach the part of the economy driving the overshoot.

• Driver: the move since July 27–28 is real-yield-led. A longer-horizon replication of the same decomposition back to March 2022 finds an even larger term-premium share (65%), and real yields have now caught up to trend real GDP growth for the first time since the 1990s.

• Peak inflation hysteria confirmed: the core CPI/PCE wedge is the widest since the early 1980s in a series going back to 1960. Strip out the stock-market-driven portion of the gap (portfolio services) and a real, AI/software-driven wedge remains.

• Equities rotated again. The healthcare/financials decoupling and the semiconductor/momentum pullback both peaked in the August 19 - 27 window and have unwound in lockstep since. These two independent dispersion trades are consistent with a single macro factor reasserting itself.

• Positioning: Pinebrook’s own closing of semi/momentum shorts and adding duration are both consistent with the data, but the duration trade is a bet that we remain in the anchored regime described above, En-Shala.

I. The Rates Regime: Fall 2023, Revisited

Rates, both nominal and real, are being described everywhere (including here!) as “back to levels last seen in fall 2023.”

Using the actual October 2023 cycle highs as the reference point rather than a vibe, three of these four series are just restating old news.

Source: US Treasury real-yield series and Bloomberg (USGGT10Y, USGGT30Y, USGG10YR, USGG30YR).

• Real 30Y has been above its fall-2023 high for twenty months.

• The only series both elevated and freshly notable is real 10Y, which crossed its own fall-2023 peak on September 10^th.

The level comparison understates how different this moment is from October 2023, because the shape of the curve is different.

Source: US Treasury real-yield series and Bloomberg

• In October 2023 the curve was inverted. Today 5% is the floor of the very long end, with a positively sloped curve underneath it.

• That change in shape, not just the coincidence of a shared 10Y print, is what makes the 2023–2024 playbook hard to run with.

• What pulled the 10Y down from 4.98% back then was the policy rate coming down with inflation.

• The current policy rate is already 175bp lower than it was that day, and the 3-month has followed it down 153bp, but the 2Y has only given back 51bp, and the 30Y is higher than its 2023 peak.

July 28 & Jackson Hole

These pages dated a rates shift on July 27 - 28, when the factor generating higher long yields changed from a term-premium story to a real-neutral-rate. Six weeks on:

• Chair Warsh’s Jackson Hole keynote produced the single largest daily RRFNR move of the window at +11bp, with term premium falling 3bp the same day.

• Warsh explicitly disavows forward guidance (a “hall-of-mirrors problem”), which should raise uncertainty compensation.

• Instead, the market traded his other message: business capex running near 9% y/y, the fastest since 2021, more than half attributable to AI, alongside inflation he called “not self-executing.”

The bond market read the growth case as credible and repriced RRFNR upward. Days later NY Fed President Williams attributed rising yields directly to AI and data-center investment, and by rate-sensitive equity sectors (Real Estate, Utilities) selling off right after the speech.

II. Is This a Regime Change? The Fork

Read more

Executive Summary

Read more

• Pinebrook generated a +21.58% return for the period January 2 through August 31, 2026, against an S&P 500 return of +12.28%, for a spread of +9.30%.

• August produced a +7.79% return against the S&P 500’s +2.62%. This was the second-best monthly alpha of the year at +5.17%.

• The result directly validates the July 13–16th repositioning. YTD risk metrics have recovered materially from the July trough: Sharpe 1.20 (up from 0.83), Sortino 1.18 (up from 0.81), downside capture 108.6% (down from 120.3%).

EQUITY CURVE VS. S&P 500

PERFORMANCE SNAPSHOT

RISK METRICS — YTD THROUGH AUGUST 31

MONTHLY RETURN BREAKDOWN

MONTHLY RISK METRICS — 2026 (PART I)

Return, Sharpe, Sortino, Beta, Alpha, Correlation by month. Each month standalone. June and July shaded. Sharpe/Sortino: green ≥2.0 amber ≥1.0 red <0

Alpha is annualized Jensen’s alpha. April’s -196.8% reflects underperformance vs beta-adjusted benchmark (beta 2.12, SPX +10.42%). Single-month alpha is sensitive to beta estimation over short windows.

MONTHLY RISK METRICS — 2026 (PART II)

Volatility, drawdown, win rate, capture ratios, best and worst days by month.

Max DD per month measured from rolling peak within that month. August downside capture of -22.7% reflects a month with very few down days for the portfolio vs. the market — directionally a strong result, but the sign artifact is a function of the portfolio falling less than the benchmark on negative SPX days.

PORTFOLIO COMMENTARY

January – July: Summary

The portfolio delivered five positive months from January through June, with March (-5.34%) and July (-5.37%) as the two down months. The H1 2026 Performance Report and the July 2026 Performance Report address each period in detail, including the concentration risk that drove June and July’s deteriorating risk metrics and the corrective repositioning executed July 13–16th.

August: +7.79%

August was the portfolio’s strongest month on a risk-adjusted basis, producing a +7.79% return against the S&P 500’s +2.62% for a +517 basis point spread. The monthly Sharpe of 6.27 and Sortino of 9.94 reflect both the strength of the return and its consistency — the portfolio won 66.7% of trading days with a best day of +2.55% and worst day of -1.01%.

The primary drivers were GDX and GDXJ (combined +$32,350 from entry across all four tranches as gold continued its run), IGV (+$32,006 from the July 16th entry), TSM (+$23,095), and the GEV and CAT short positions (+$12,822 and +$13,266 respectively) as industrial machinery underperformed.

EWJ short (combined -3,980 shares) remained the primary drag, down approximately -$24,000 for the month as Japanese equities continued to strengthen. The DXJ long partially offset this through the pairs structure, but the net Japan position was a headwind. KWEB long was also a small drag (-$2,775 on the 3,384 share position closed August 21st) as Chinese internet names softened late in the month.

Seven positions were closed on August 21st: SPHR (+$14,115), LYV (+$7,778), FEZ (+$3,654), EPOL (+$11,819), FXI (+$3,543), MCHI (+$2,584), and KWEB (-$2,775). Combined realized gain from these closures: +$40,718. The remaining book heading into September is more concentrated in the gold miners, the Japan pair, IGV, TSM, and the GEV/CAT industrial short basket.

ON THE RISK METRICS

The monthly metrics matrix tells the August story directly. The two-month anomaly of June (Sharpe 0.59) and July (-3.37) has been followed by the portfolio’s strongest risk-adjusted month of the year. The July report stated that whether the repositioning was sufficient would be visible in the August numbers. It is.

YTD metrics have recovered across every dimension: Sharpe from 0.83 to 1.20, Sortino from 0.81 to 1.18, downside capture from 120.3% to 108.6%, annualized alpha from 8.02% to 14.84%. The maximum drawdown of -10.29% (July 23rd) remains in the YTD series and will continue to roll off as the portfolio builds on the August result. Beta has compressed from 1.13 to 1.08 and correlation from 0.60 to 0.58, consistent with the repositioned book running less single-factor exposure than the June–July peak.

TRACK RECORD

Portfolio performance since inception in August 2024.

For those that don’t want to use Discord, the Substack chat feature will also be used, as well as long form emails when appropriate.

Here’s how to join the Pinebrook Discord, it takes about 2 minutes:

1. Join the Server.

2. Once you’re in, fill out this short form with the email you subscribed with and your Discord username on this Google Form.

3. You’ll be automatically added to the Core channels within a few minutes. No need to wait on me.

A few things to know:

• The old X private feed will be shut down on September 15, 2026. If you want uninterrupted access, get set up on Discord before then.

• Going forward, your Discord access is tied directly to your subscription — it activates and deactivates automatically, so there’s nothing to manage on your end (or mine!).

• If you run into any trouble linking your account, just reply to this email and I’ll sort it out directly.

Looking forward to having you in there!

PS. I expect to have the blotter and dashboard up by the time the X account is shot in the head on September 15^th.

Last week’s note on August 20th established that the rise in long yields shifted on July 27, 2026, from a real-neutral-rate story to a term-premium story, concentrated almost entirely at the long end. That note answered the why of the movement.

Today’s note seeks to tackle a bigger question: what kind of world produces that move?

• At the risk of giving it away, Pinebrook’s view is that the U.S. economy is currently undergoing an inflationary boom expansion.

• While many market participants, too unborn to even remember disco and Studio 54, much less a man on the moon, have been conditioned by 40 years of disinflation and fifteen years of low-rate orthodoxy, the bond market is simply along for the ride and pricing the above.

• Thus, the dog is wagging the tail and not the other way around as the so-called bond vigilantes would have you believe.

Party like its 1969, not 1999.

I. The Template: Guns and Butter, Twice

Wars are expensive. A government funding one typically faces a choice: raise taxes and cut domestic spending (guns OR butter) or run both simultaneously on the national credit card (guns AND butter).

President Johnson (“LBJ”) made the second choice, financing the Vietnam War while expanding the Great Society social safety net.

Fiscal and Inflation Scorecard, 1964 - 1969

The current and prior administrations have made the same structural choice, with one governing difference: LBJ’s butter went to the poor and working class; this cycle’s has gone to asset holders.

That distributional difference matters for who benefits (Lower-K wage earners vs. Upper-K owners of capital), but not for the causal mechanism that moves rates: booming nominal GDP (nGDP).

When the Current Clock Actually Starts

Using the current Iran conflict as a starting point for a LBJ Guns & Butter comparison era is lazy partisan thinking, as it undercounts how long the Guns & Butter agenda has been running.

A more appropriate baseline is January 2020. This is a five-year plus structural story.

• The war (butter) on Covid, marshalled the fiscal and monetary resources of the Federal Government and resulted in WWII-levels of fiscal spend and monetary accommodation.

o The CARES Act of 2020 under President Trump was a $2.2 trillion spending bill.

o The American Rescue Plan Act of 2021 under President Biden was a $1.9 trillion spending bill.

• Tariffs are likewise on the guns side of the fiscal ledger: a national consumption tax, imposed on households and importers, that show up in the price data the same way a war bill shows up in the outlay data.

• The OBBB reconciliation bill of 2025 under President Trump was more lard butter: upper-income tax cuts plus maintained domestic spending, financed on the credit card rather than through offsetting revenue cuts is identical to how LBJ financed the Great Society, just aimed at a different income cohort.

The IEEPA Tariffs Were Struck Down

On February 20, 2026, the Supreme Court ruled 6-3 that the tariffs imposed under the International Emergency Economic Powers Act exceeded presidential authority. Taxation power belongs to Congress, not the executive acting unilaterally under emergency powers.

• The government had collected an estimated $130–175 billion under that regime and is now refunding it.

• As of mid-August, roughly $100 billion (60%) has gone back out to importers.

The Whole Hog, 2020–2026

*10Y and CPI figures are calendar/fiscal-year-end approximations compiled from public trackers. Federal outlays are actual/reported FY totals; FY26 is a partial year (through July) and not directly comparable to the full-year columns beside it.

Highlights:

• 2020 → 2024: outlays rose from $6.55T to $6.75T, but the path wasn’t linear.

• Spending peaked near $6.82T in the depths of pandemic relief (FY21), fell back to $6.27T as emergency programs lapsed (FY22), then climbed again through FY23–24 even as the emergency itself was long over.

• The post-2020 baseline never returned to pre-COVID norms, regardless of who sat behind the Resolute Desk.

• 2025 → 2026 YTD (the current leg): outlays continue climbing from an already-elevated $7.01T base, now layered with tariffs, OBBB, and the Iran war.

The structural change that matters is not the spending itself, but its context:

• An economy already near full employment with little to no labor market slack.

• A wall of money hit an economy with supply chain distortions that were magnified by the sheer amount of cash that was put to work to bid prices up and clear tight markets.

What separates an inflationary boom from countercyclical stimulus is an overheating of the labor market instead of simply protecting it.

• U3 moonshot from 3.5% in February 2020 to 14.7% in April 2020.

• U3 ended the 2020 calendar year at 6.7%.

• By calendar-end 2021, U3 was at 3.9% - back to full employment.

• Monetary policy did not respond until Spring of 2022.

• Fiscal restraint did not ensue despite the return to full employment.

Cumulative Federal Deficit Fy20 Fy26

26.9KB ∙ PDF file

Download

Download

The trajectory and outcome of the Iran War is the single biggest risk to this expansion remaining a clean inflationary boom rather than tipping into a stagflationary recession.

The reason is that the escalation could accelerate inflation faster than the tariff-and-tax-cut baseline alone would.

II. The Receipts

• National debt crossed $40 trillion last week.

• Excluding ~$8T in intragovernmental holdings, Treasury debt held by the public is now roughly 100% of nominal GDP.

• Interest expense is approaching 1.0% of GDP, up from a 0.6–0.7% pre-pandemic norm sustained for roughly two decades.

• The Reagan-era peak was 1.25%.

• The 2026 YTD deficit already exceeds every full year between 2023 and 2025, inclusive.

• Debt held by the public: $31.9T vs. $29.4T a year prior.

• Notably, Petersen attributes the increase mainly to net interest (+$91B), Social Security (+$71B, COLA-driven), and Medicare (+$65B) — the entitlement-and-interest core of the budget, not tariffs or the war directly.

• The counterweight: total nonfinancial debt-to-GDP (public plus private) has been broadly stable since 2010.

• The rise in government leverage has been partly, not fully, offset by private-sector deleveraging.

III. Adjudicating the Rival Explanations

Read more

Gold is indeed shining again, after being taken to the woodshed and left for dead this spring following its parabolic move into the 3-sigma area in January.

In fact, gold round-tripped its up move back down into the negative 2-sigma territory, with a final bottom on July 16, 2026.

Now with financial media catching up to the trade +16% off the July bottom, some readers would like to know if the bounce should be faded (sold), let to run (maintain position), or added.

Below is an update on where things stand given the levels used in the analytical framework of the evaluation.

Read more