• Base case is for a hawkish-hold on the policy rate.

• The soft inflation data for June provides the cover for not hiking in July.

• The committee will inject a stronger tightening bias and tee-up up a hike for Q3.

• The resumption of hostilities in the Middle East, diplomatic or kinetic, argues for more supply-chain driven inflation.

Since Fed Chair Warsh’s June press conference, these pages have argued that future FOMC meetings would have little left to reveal. In part due to Chair Warsh disdain for forward guidance, but also in part because the hawkish reaction function was already out in the open.

That thesis holds for July 28 – 29. But this week the noise has multiplied rather than thinned.

• A live Fed decision with no fresh dot plot.

• An earnings season whose headline growth rate is partly a GAAP accounting artifact.

• The two largest hyperscalers on earth reporting the same afternoon Chair Warsh speaks.

None of that is the signal.

The Signal: RRFNR Has Round-Tripped

Read more

These pages will skip the traditional “sell in May and go away” seasonal platitudes in favor of teasing out a forward-looking signal from the price action.

The chop is not a total surprise.

• Summer was flagged as a window of weakness back in the dark days of the start of the Hormuz energy shock.

• August was flagged as the month when inflation and growth pressures emanating from the Hormuz shock would fully manifest themselves and get digested by the market.

• Instead of index level volatility, the eye of the storm has been in market internals at the sector level, which have been experienced a violent rotation from the high-beta semiconductor momentum trade to other sectors.

MTUM is down -8.43% MTD for the month of July. Some of its bigger components are faring worse:

• MU is down -18.28% MTD.

• CAT is down -15.72% MTD

• GEV is down -12.925% MTD.

Meanwhile, SMH, the semiconductor ETF, is down -12.831% MTD.

The salient summer question is this: is this the end of the unwind or does it have more to go?

Read more

• June CPI (July 14) delivered the first real Core CPI miss in months. Most of it doesn’t matter for PCE.

• June PPI (July 15) delivered the part that does matter: airfares and portfolio management came in soft, and that’s the real impact on the core PCE call below.

• A July hike is off the table, but these pages have front run that.

• The house call is for a December hike, but September/October are still alive and for now genuinely data-dependent rather than a foregone conclusion.

Read more

Read more

Here is what the performance scorecard looks like since June 2^nd, the day XLK peaked, through July 10^th.

The rotation is no longer a speculative theory.

• There is a nearly 1,600 basis points of spread between the best and worst sectors since the semi peak (XLV vs. XLK).

• That is not noise. It is a shift within the current bull market.

In the June 8th note, the 5-sector 30-day rolling correlation average vs. the S&P500 stood at 0.491, which was described as -3.58σ below the post-2018 mean. It was flagged as one of the most extreme readings in the 27-year dataset outside specific market dislocations.

There was an underestimation in how much further it would go.

As of close on July 10th, the composite correlation average stands at 0.456.

Read more

Here is a trade list and accompanying rationale.

Portfolio Positioning: Executing the Rotation

The correlation data and research converge on a set of portfolio moves that express the rotation thesis.

Read more

• For the month of June 2026, Pinebrook generated a return of 1.77% versus a loss of -1.06% for the S&P500; an outperformance spread of 284-basis points.

• YTD for H1 2026, Pinebrook has generated a return of +19.19%, against a S&P 500 return of +9.55% — a spread of 964-basis points.

• On a risk-adjusted basis, the portfolio produced a Sharpe Ratio of 1.37 and a Sortino Ratio of 1.39.

• The H1 maximum drawdown was -9.56%, and the peak-to-trough window ran from May 29th through June 10th, with the book recovering to new highs by mid-June.

• The local peak to trough drawdown for the month of June took place between June 2^nd and June 10^th.

• Beta of 1.20 and correlation of 0.63 reflect a concentrated, directional book with meaningful but selective market exposure.

• Upside capture of 136.6% and downside capture of 116.8% — combined with an annualized alpha of +19.22% — reflect a portfolio consistently generating returns independent of broad market direction.

• The last two points reflect a deterioration in risk metrics from the prior month, discussed below.

PERFORMANCE SNAPSHOT

RISK METRICS

MONTHLY RETURN BREAKDOWN

COMMENTARY

January – February: +6.28% / +1.84%

The book opened the year decisively, posting +6.28% in January and +1.84% in February against a flat-to-negative S&P 500. The primary driver in January was a concentrated position in Korean equities via EWY — entered at $104.27 and closed January 30th at $123.96, an 18.9% return on the position. DXJ, capturing yen-weakening dynamics, contributed an 8.9% gain and was exited in mid-February. Silver micro-futures provided an additional non-correlated contribution in late January and early February, contributing approximately $40,000 in combined PNL. The cumulative effect was a +8.26% portfolio gain through the first two months against the S&P 500’s +0.48%.

March: −5.34%

March was the book’s weakest month, declining -5.34% against the S&P 500’s -5.09%, a -25 basis point negative spread. A cluster of position unwinds — KRE, WNC, RSP, EWU, and DXJ — moved against the book simultaneously during a period of broad risk-off positioning driven by tariff uncertainty. The near-parity with benchmark performance reflects the portfolio’s genuine equity character; broad sell-offs correlate across the book.

April – Ceasefire Announcement: +4.02%

April produced a net gain of +4.02% despite containing the period’s maximum drawdown event. On April 7th the portfolio reached its intra-period trough, driven by GDX/GDXJ long exposure and crude oil futures positioning ahead of the Ceasefire Announcement. April 8th produced the period’s best single day at +4.88% as ES and NQ futures captured the market reversal. The April 9–10 cluster — long-biased futures and long EEM — produced additional gains as global equities recovered sharply. The S&P 500’s +10.42% headline April gain was driven primarily by the post-announcement snapback; Pinebrook’s +4.02% reflects entry positioning that was partially caught in the initial drawdown.

May: +9.89%

May was the strongest single month at +9.89%, outperforming the S&P 500’s +5.15% by 474 basis points. Semiconductor exposure via SMH, clean energy positions (QCLN, PBW), and broad emerging market participation through EMXC were the primary contributors. Short positions in EIDO and INDA also contributed positively as those markets underperformed.

June: +1.77%

June produced a +1.77% gain against the S&P 500’s -1.06%, a +284 basis point spread and the second-best monthly alpha of the year. The month was characterized by strong cross-asset dispersion: open long positions in EWY, TSM, SPHR, and SMH drove gains, while China shorts (MCHI, FXI, KWEB) contributed a combined $26,000 as Chinese equities softened. XLE, OIH, QCLN, and PBW were closed mid-month on June 18th. The ESM6 carry position was closed June 15th at a gain, with the ESU6 roll entered same day finishing the month modestly below entry.

The month’s defining session was June 5th, which produced the period’s largest single-day drawdown. The catalyst was a strong May jobs report — 172,000 jobs added with 93,000 in upward revisions — that sharply repriced rate-cut expectations and triggered an aggressive selloff in rate-sensitive technology and semiconductor names. XLK fell 6.66% on the session; EWY fell 14.1% as Korean semiconductor names bore the brunt of AVGO-driven contagion. The portfolio’s concentrated exposure to that trade — EWY, SMH, and ESM6 futures each representing a variant of the same AI/semiconductor position — amplified the impact.

The session is best read as a rotation event rather than a risk-off episode. XLV and XLF both closed positive on the day the S&P fell 2.58%, and three of the five major S&P sectors finished the week in positive territory. Capital was being reassigned, not withdrawn. Consistent with that reading, the portfolio recovered the bulk of the June 5th loss within two sessions and finished the month with a positive return and the second-widest monthly spread against the benchmark in 2026.

On the risk metrics: the YTD Sharpe and Sortino ratios reflect the full January–June period and show some June deterioration versus the May report.

The portfolio ran a more concentrated, directional book in the second quarter — EWY, TSM, SMH, and EMXC in size simultaneously — which is a genuine increase in thematic concentration risk relative to the more diversified January–February positioning. This will be addressed going forward.

GETTING THE ECONOMICS WRONG AND THE TRADE RIGHT

The macro call that framed Q2 deserves a “come to Jesus moment”, and my own research timeline provides the receipts.

• Beginning in late March, Pinebrook’s published work made a high-conviction case for near-zero Q2 GDP growth, a prolonged Hormuz closure, and physical Brent crude reaching $150–200 per barrel by end-June.

• The thesis was a 900-day inventory deficit, a $150 billion regressive energy tax on the US consumer, and a ceasefire that triage-solved the tail without reopening the Strait at meaningful scale. The call for zero GDP was explicit.

• It was the wrong economic call. Q2 GDP came in at +2.7%. Brent is back near pre-conflict levels. The Strait ran longer than the one-month kinetic baseline assumed — roughly three and a half to four months of acute disruption — and growth smashed the zero forecast.

• The July 4th economic update stated directly: the miss was “very wrong” on the GDP call and “very correct” on the view that the US would outperform its peers and that the AI capex cycle would hold long enough to bridge the wedge between shock and normalization.

• The underestimate was in the offsetting forces — the restocking cycle, labor hoarding, and above all the AI capex cycle — relative to the mechanical GDP drag modeled from the supply shock.

The PNL, however, had a different experience from the macro forecast.

• Q2 produced three consecutive positive months — April +4.02%, May +9.89%, June +1.77% — with no down months in the quarter. The reason is sequencing and the willingness to change the view when the evidence changed.

• On the evening of April 7th, as markets absorbed the ceasefire announcement, Pinebrook pivoted and got long risk.

• The repositioning that followed — into Korean semiconductors, Taiwan, broad semiconductors, and EM — was the book’s expression of a probability shift.

• The worst-case scenario had been taken off the table, which concentrated forward probability toward the AI earnings cycle that equity markets were already pricing two years out.

• The central thesis that ran through all of the published research — that the US would outperform the rest of the world through the shock, supported by reserve currency status, energy self-sufficiency, and the AI capex cycle as a structural floor — held throughout and remains the organizing framework for H2.

• Getting the economics wrong and the trade right is a comfortable place to be because PNL > cute stories.

• The discipline of updating when the evidence changes is part of the process, and Q2’s results are the outcome.

TRACK RECORD

Portfolio performance since inception in August 2024.

Read more

• Growth into H2 2026 rests on three pillars: a consumer spending story that is nominally strong but real-income weak; an AI capex cycle that is now visibly bleeding into the physical economy; and a labor market showing the first hard signs of reacceleration after a year of stagnation.

• Pinebrook’s focus is on the compositional undercurrents that the headline numbers don’t show, and the Fed’s reaction function to the forward evolution of the data.

• Historically, cycles that looked like this — long leading indicators flashing yellow, real income negative, headline data still fine — have been rescued by two things: decelerating producer prices and loose fiscal policy.

• June’s jobs report, which was weak on payrolls but strong on unemployment, bought the Fed some time: the rate hike that was getting priced for October has been kicked to December.

• Pinebrook’s 10-year short position was closed at a loss — the first rates loss of the year.

• The house view on the rates complex is neutral pending clarification from the July jobs report and the August Supercore read.

Where We Left Off

Back in April of this year, we were in phase one of the energy supply shock timeline, when the ceasefire suddenly removed the worst-case tail and concentrated probabilities toward the AI earnings cycle thesis and drove the equity recovery.

• Productivity had not collapsed, and labor hoarding kept U3 stable while hours worked and soft survey sentiment deteriorated.

• The sectoral sequencing that was laid out — transportation and logistics first, manufacturing second, construction third, consumer goods fourth, services last — had begun its first stage.

• The key regime question — temporary price volatility or persistent supply instability — was the unknown.

What was known was that the U.S. was entering the shock from a position of relative strength due to its reserve currency status, energy self-sufficiency, and an AI investment boom standing in for the late-90s tech cycle.

Thus, Pinebrook’s very wrong call for zero GDP growth output was offset by the very correct view that the U.S. comes out ahead of its peers and that the AI capex cycle would hold long enough to bridge the wedge between shock and normalization.

That thesis has aged well.

• The Iran MOU was signed on June 17 and closed out the acute phase of the shock.

• Brent is back to pre-conflict levels near $72.

• The left-tail growth risk has been shot in the head, and U.S. assets outperformed RoW assets on a 100-day rolling return basis.

For those watching at home, the math holds the same on an absolute price return basis (April 24 – June 30):

While we got the trade right (what really matters), we got the economics wrong.

• The Strait closed in late February; the Iran MOU wasn’t signed until June 17 — roughly three and a half to four months of acute disruption, longer than the one-month kinetic baseline the March 30th note built its GDP math around.

• The shock ran longer than assumed, not shorter, and growth still came in near 2.7%.

• That means the miss was an underestimate of the offsetting forces — the restocking cycle, labor hoarding, and above all the AI capex cycle — relative to the mechanical GDP drag modeled from the supply shock.

Looking Forward

Read more

Read more

The dot plot shifted hawkishly at the June FOMC.

Read more

Read more

Read more

Before doing so, please recall the January 2026 note on SLV, when peak insanity bid the silver ETF up between six and seven standard deviations above the historic 100-day rolling return SLV-SPY spread of 76-baisis points.

Read more

The core argument was straightforward: gold’s financialization via the GLD ETF in 2004 had transformed it from a barbarous relic into a fully fungible, 21st-century financial instrument. Central bank accumulation, an insatiable retail bid, and gold’s evolving hybrid risk-on/risk-off correlation profile made it a structurally sound long.

• At the time of that August note, the 100-day rolling GLD-SPX spread sat at 5.09%, having compressed sharply from 41.24% at the April 2025 tariff-tantrum peak.

• The S&P 500/gold ratio was 1.86x against a long-term average of 1.6x, implying a fair-value gold price of approximately $3,900/oz.

By September 5th, a note titled New Trade Discussion added gold miners for convexity. The rationale: junior miners in the GDXJ ETF were operationally leveraged to the gold cycle, reporting record revenues, record earnings, record operating cash flow, and a record average profit per ounce of $1,900 against all-in sustaining costs. Miners had delivered 70–90% year-over-year earnings growth for eight consecutive quarters. Pinebrook preferred GDXJ over GDX for its superior growth profile.

Then came October 20^th. Gold had risen 132% from its October 2, 2023, bull-market birthday, the strongest absolute advance in the history of the yellow metal.

The GLD ETF had closed within a whisker of being 2-standard deviations overbought relative to the S&P 500 on a one-calendar-year lookback, reaching 47.22% versus a 2-sigma threshold of 48.19%.

The same note sounded caution. The average drawdown across the prior gold bull cycles was -16.3%. A correction was less of a matter of if and more of when and how deep.

The answer came on January 28, 2026, when gold peaked at $5,417.21/oz, having surged a parabolic 25.1% in January.

• At peak, the GLD/SPX spread was stretched to 79.69% above its 200-day rolling return, the most overbought reading since March 1980.

• On a one-year rolling basis, the GLD-SPX rolling return spread was more than 3 standard deviations above the mean, a level unseen in 20 years of data.

In a note titled Miners, published February 2, 2026, Pinebrook got short the miners.

Five Months of Reckoning

Gold’s drawdown from the January 28th peak has been ugly. As of June 11, 2026, gold spot trades at $4,212.26/oz, a decline of 22.24% from the January 28^th peak of $5,417.21/oz.

The June 11, 2026, intra-day low of $4024.01/oz. represents a peak-to-trough drawdown of -25.72%, exceeding every prior correction in the post-2006 gold super-cycle, where the historical average peak-to-trough decline across 11 prior episodes was -16.30%.

Read more

The market did not panic. It rotated.

Friday’s catalyst was the May jobs report: 172,000 jobs added, with net upward revisions of 93,000 to prior months. The three-month average rose sharply to 188,000. It was the third strong report in a row.

A strong jobs print with upward revisions does one thing above all else in the current environment: it raises the probability of a hike not only of one hike, but of multiple hikes this year.

The 2-year Treasury, already at 4.04% heading into the number, and gapping up to 4.14%, tells the story.

The sector breakdown of Friday’s price action is worth examining in detail:

In a genuine risk-off event, all six of those rows are red. The fact that XLV and XLF held positive on a day the index dropped 2.58% tells you this was a Tech-specific repricing, not a macro panic.

The weekly numbers confirm it:

• Health Care rose 2.37%.

• Financials rose 1.4%.

• Consumer Discretionary fell 4.97%.

• Industrials were up .61%

• IT fell 5.61%.

Three out of five sectors were up on the week. The market is reassigning capital.

For a sector ETF of XLK’s size and liquidity, -6.66% in a single session is a huge cyclical move. The AVGO contagion, the payrolls catalyst, and the rate repricing all hit simultaneously. Whether Friday was capitulation or the beginning of a more sustained drawdown is the central question heading into the week.

• XLK has retraced a meaningful portion of its 55.45% rally from the March 30th trough to the June 2nd peak.

• Now at a 41.41% advance as of June 5th.

• One of the most important data points this week received the least attention:

• Year-to-date through Friday, the Roundhill Magnificent Seven ETF (MAGS) is up just 0.83%.

• The S&P 500 excluding the Mag-7 (XMAG) is up 9.87% YTD.

• The cap-weighted S&P 500 ETF (SPY) is up 8.16% YTD.

• The equal weighted S&P ETF (RSP) is up 8.49%.

The other 493 stocks in the S&P 500 are collectively outperforming the seven largest companies by roughly 900 basis points year-to-date and helps to validate the rotation call.

Does This Look Like 1999?

A reader posed this directly: what did this look like in the 1999-2001 rotation? Does some level of concentration necessitate rotation equals selloff?

We now have 27 years of daily data on all five sector ETFs — XLV, XLF, XLY, XLI, and XLK (all launched December 22, 1998) — which allows a proper apples-to-apples comparison. The answer requires going sector by sector, not relying on the composite.

Chart 1: 30-day average rolling correlations of all five sectors vs. SPY, full 27-year history. Note the September 2000 trough at 0.188 — visually in a different universe from Friday’s 0.474.

The two eras share a striking structural resemblance at the regime level.

• The dot-com full cycle (Jan 1999 – Oct 2002) ran a mean 5-sector correlation of 0.742.

• The current cycle (Jan 2024 – Jun 2026) is running at 0.725.

Within rounding, they share the same low-correlation environment. The frequency of sub-0.60 readings is also nearly identical: 11.8% of trading days in the dot-com window, 10.5% in the current window. These are the only two sustained low-correlation regimes in the 27-year dataset.

While the composite average looks similar between the two eras, the individual sector lines tell a completely different story. In 1999-2000, all five sectors were dispersing from the index simultaneously — a broad market-wide phenomenon driven by investors making confident sector-level bets across the entire index.

Today, three of the five sector lines are clustered relatively tightly. The anomaly is concentrated in two sectors: XLV and XLF.

Chart 2: 30-day rolling correlations for each of the five sectors vs. SPY, full 27-year history. The dot-com era (amber) and current cycle (blue) are the only two sustained low-correlation regimes.

The Structural Differences — Four of Them

1. Sigma extremity. Against the dot-com window’s own mean and standard deviation, today’s composite reading of 0.474 is -2.51σ. The September 27, 2000 trough of 0.188 was -4.64σ against that same window — and against data known in real time at that moment, approximately -4.5 to -5σ. The chart makes this point better than any number: the September 2000 collapse is nearly off the bottom of the page. We are not there. We are not even close to there.

2. Source of the low correlation. In 1999-2000, sector mean correlations ranged from 0.615 (Industrials) to 0.778 (Tech) — a tight cluster, all sectors participating. Today, XLK at 0.878 is tightly fused to the index. The low-correlation reading is driven almost entirely by XLV and XLF. Strip those two out and the remaining three sectors are running at normal-to-elevated correlations. This is a two-sector anomaly, not a market-wide one.

3. Trajectory pattern. The dot-com regime produced a single catastrophic break — a near-vertical collapse to 0.188 in September 2000, followed by a snap recovery within weeks. The current cycle has produced three distinct dips below 0.60: July 2024 (monthly average 0.479), February 2025 (0.593), and now June 2026 (0.491). Each has recovered. But the recoveries are getting shallower and the floor is gradually descending. Repeated shallow dips suggest rotation. A single violent break suggests dislocation.

4. Market return context. SPY price return during the dot-com full cycle (Jan 2, 1999 – October 9, 2002): -36.66%. XLK: -64.5%. SPY total return in the current cycle (January 2, 2024 – June 5, 2026): +56.04%. XLK: +92.38%. Same low-correlation regime signature. Completely opposite market outcomes. The dot-com bust was caused by valuations completely detached from earnings reality. The AI trade today is led by companies with real, growing, massive earnings.

Direct answer to the reader’s question:

Read more

• Understanding the Business and Its Valuation

• A Very New Story Dressed in Old Clothes

• Tesla as Benchmark

• Index Inclusion: Its an AI world

• Sector Classification

Read more

• Pinebrook generated a +17.12% net return for the period January 2 through May 29, 2026, against an S&P 500 return of +10.73% over the same period — a spread of +6.39%.

• On a risk-adjusted basis, the portfolio produced a Sharpe Ratio of 2.07 and a Sortino Ratio of 2.31.

• The Sortino exceeding the Sharpe — consistent with prior periods — confirms that realized volatility is skewed to the upside.

• Maximum drawdown was contained at -6.56%, reached on April 7, 2026, during the war ceasefire, before recovering sharply.

• Upside participation was strong at 109.1% capture, while downside exposure was meaningfully reduced at 88.2%capture. A beta of 0.82 and SPX correlation of 0.59 reflect a portfolio with market exposure calibrated to opportunity rather than passive index tracking.

• The updated blotter and dashboard, through close of May 29, 2026, can be found here.

January – February: +6.28% / +1.84%

The portfolio opened the year decisively, posting gains of +6.28% in January and +1.84% in February against a flat-to-negative S&P 500. The primary driver in January was a concentrated position in Korean equities via EWY — entered in early January at $104.27 and closed January 30th at $123.96, a +18.9% return on the position. DXJ, capturing yen-weakening dynamics, contributed an 8.9% gain and was exited in mid-February. Silver short positions provided an additional non-correlated contribution in late January and early February. The combined effect was a +8.26% cumulative portfolio gain through the first two months against the S&P 500’s +0.48%, producing 778 basis points of spread.

March: -5.34%

March was the portfolio’s weakest month, with a decline of -5.34% against the S&P 500’s -5.09%, producing a modest negative spread of -25 basis points. A cluster of position unwinds — KRE, WNC, RSP, EWU, and DXJ — all moved against the book simultaneously during a period of broad risk-off positioning driven by war uncertainty.

April: +4.02%

April contained the period’s most significant single-event trading sequence. On April 7th the portfolio reached its maximum drawdown of -6.56%, driven by GDX/GDXJ long exposure and crude oil futures positioning ahead of the ceasefire announcement. The following session on April 8th produced the period’s best single day at +4.88%, as ES and NQ futures positions captured the market reversal. The April 9th–10th cluster — long-biased futures and long EEM — produced additional gains as global equities recovered sharply. Pinebrook’s April return of +4.02% reflects entry positioning that was partially caught in the initial drawdown before the recovery.

May: +9.89%

May was the strongest month of the 2026 year-to-date period at +9.89%, outperforming the S&P 500’s +5.15% by 474 basis points. Semiconductor exposure via SMH, reopened clean energy positions (QCLN, PBW), and broad emerging market participation through EMXC were the primary contributors. The closure of short Indonesia (EIDO) and short India (INDA) positions also contributed positively as those markets underperformed. Open positions at period close include meaningful long exposure to EWY, TSM, and ESM6 futures, along with a diversified book entered in the final week of May.

The following table places 2026 YTD performance within the portfolio’s full history since inception in August 2024.

It is fine right now in the same way that the Mediterranean was calm in the straits between Scylla and Charybdis until Odysseus had to sail through them. The calm is real. The dangers flanking it are equally real.

The purpose of this note is to identify what investors need to know so that they come out the other side of the strait with their ship and PNL intact.

I. The Market: Not FOMO

Read more

Read more