The reasoning was simple. Whoever runs the Fed decides how cheap or expensive borrowing will be for the whole economy, and that single decision ripples into the dollar, gold, Bitcoin, smaller US companies, the giant tech stocks — basically everything you’d want to invest in.

Nine days later, on May 13, the Senate confirmed Warsh by a vote of 54 to 45. It was the closest Fed Chair confirmation in modern American history.

Powell stepped off the chair on May 16. The event I called the pivot of the quarter is now in effect, six weeks earlier than I expected.

That speed matters, because it stress-tests the framework in real time. I thought we’d have until July before the new Fed era began shaping the data. Instead it began the moment Warsh’s vote cleared, the Iran war re-escalated on May 11 (oil briefly back above $100 a barrel, Netanyahu warning that “the conflict is not over”), and four of my twelve Q2 predictions are visibly off-trajectory.

This is when a newsletter has to earn its keep.

So here’s what I’m doing: a public mid-quarter scorecard, with directional verdicts on every prediction, what I got right, what I got wrong, and what I’m updating.

Not in August at the next quarterly. Now.

Because writing predictions in public is easy.

Owning them while they’re bleeding is the part that actually matters.

The two-week read on Issue #1

These aren’t final verdicts — Q2 doesn’t close until June 30. They’re directional reads as of the close on May 18.

At this stage, which way the needle moved matters more than whether the target was hit.

═══════════════════════════════════════════════════════════════
        ISSUE #1 — Q2 PREDICTIONS · MID-QUARTER VERDICT
═══════════════════════════════════════════════════════════════

  TIER 1 — HIGH RISK / HIGH REWARD
  ─────────────────────────────────
  #1  Big tech AI stocks crash -25 to -40%   ✗  FADING
       Microsoft, Google, Amazon and Meta confirmed
       roughly $700 billion of AI investment for 2026 —
       higher than expected, not cut. The crash trigger
       I was waiting for hasn't fired.

  #2  Uranium price to $130-150/lb            →  IN PLAY
       Spot price at $86.55, long-term utility contracts
       at $90 (highest since 2008). Going the right way,
       just not as fast as needed.

  #3  Yen rallies hard against the dollar     ✗  WRONG (so far)
       Yen is at 158.90 per dollar — the wrong direction.
       The Iran war pushed investors back into dollars,
       and the Fed is now expected to raise rates instead
       of cutting them. Full reset of this thesis below.

  #4  Private credit / mid-market lender crisis  →  IN PLAY
       No big lender has had to freeze withdrawals yet.
       Credit spreads are widening though, and the wave
       of corporate debt that needs refinancing is still
       ahead of us — just not in this 14-day window.

  TIER 2 — MEDIUM RISK / STRONG REWARD
  ─────────────────────────────────────
  #5  Copper price to $13,000 per tonne       →  IN PLAY
       LME at roughly $10,000/t. Goldman thinks supply
       will catch up; JPMorgan still sees $12,000+. The
       call is delayed, not dead.

  #6  European defense stocks lead the market ✗  BADLY WRONG (so far)
       The main European defense ETF (EUAD) is down 4%
       this year, and the wider sector has lost as much
       as 25% in some measurements. Rheinmetall fell 9%
       in a single day on May 8. Full rework below — this
       is the prediction that's hurting most.

  #7  US dollar weakens, DXY index below 95   →  IN PLAY
       Currently at 97.84 (peaked at 99.35 when Iran
       flared up). The trend is right, but the Iran war
       interrupted the timing.

  #8  Gold to $5,000 an ounce                 ✓  ALREADY HIT (then retraced)
       Hit $5,589 in January — well above target. Now
       back to $4,694. Target was met; thesis now shifts
       to whether gold defends the new floor.

  #9  Bitcoin breaks $130,000                 →  IN PLAY
       Trading around $93,816 on the latest rally. ETF
       inflows are back ($700 million in a single week).
       Still well below target, but the trend is intact.

  TIER 3 — LOWER RISK / MODERATE REWARD
  ──────────────────────────────────────
  #10 Russell 2000 (small caps) beats S&P     ✗  FADING
       Russell up 14% year-to-date, S&P up 17%. So small
       caps are *underperforming*, not leading. Goldman
       has gone publicly negative on the trade.

  #11 Chinese stocks (MSCI China) up 15-25%   →  IN PLAY
       Tracking. The Iran shock pushed money into safe
       assets in April, slowing the rally.

  #12 US utilities beat S&P by 5+ points      ✓  ON TRACK
       The thesis (data centers need massive electricity)
       was reinforced by the hyperscaler earnings. The
       trade is working quietly.

═══════════════════════════════════════════════════════════════

  Score so far: 2 hit / on track · 5 in play · 4 wrong-so-far · 1 fading

A 17% hit rate at the two-week mark. Not great. But here’s why I’m publishing it anyway: the predictions that are wrong are wrong for identifiable reasons — the Iran war reshuffled the deck, and the Fed is now expected to raise rates instead of cutting them.

The underlying framework didn’t break.

The order in which things were supposed to happen did.

That’s a fixable problem, not a fundamental one.

The full update on each of the eleven still-active calls, the three new predictions I’m adding for the back half of Q2 in light of Warsh’s arrival, and the specific stocks, options, and instruments I’m using to express each view — that’s behind the wall.

But before the paywall, I want to do one thing I think nobody else in this space does: I’m going to walk you through my worst-performing prediction in full, for free. Not the spin.

The whole analysis, what I missed, and what I’m doing now. Because if you’re considering subscribing, you should see exactly what depth looks like — including on the trades that hurt.

Revisiting Prediction #6 — European Defense

The pain. EUAD, the main exchange-traded fund that tracks European defense companies (Rheinmetall, BAE Systems, Leonardo, Thales, and similar names), opened 2026 at roughly $46 a share. It closed May 16 at $41.

That’s an 11% decline at the same time the S&P 500 is up 17%, meaning Europe defense has lagged the broader market by about 28 percentage points. The pain in individual stocks is sharper: Rheinmetall dropped 9% in a single session on May 8, hitting its lowest level in a year.

When you call a prediction with 70% confidence and it loses 28 points to the market in four months, you owe your readers a precise account of why. So here it is.

What I missed. Three things.

First, I didn’t take seriously enough how much investors would pay attention to every rumor of a Russia-Ukraine ceasefire.

Every time President Trump suggests a framework, European defense stocks lose 4 to 6 percent. I knew that risk existed. I didn’t size it correctly. A “70% confidence” call should have been closer to 55% given how loud the political noise has been.

Second, I underestimated the way the Iran war (which started February 28) pulled defense investor money toward American companies — Raytheon, Lockheed Martin, General Dynamics — and away from European names.

Those US contractors outperformed their European peers by 8 to 11 percentage points in the first quarter alone.

The story “Europe is rearming” got drowned out by the more immediate story “America is fighting in the Middle East right now.”

Third, and most importantly, I confused signed contracts with stock prices. Germany’s €108 billion defense budget for 2026 is real. The €50 billion in procurement approved last December is real. McKinsey’s projection of roughly €800 billion in annual European defense spending by 2030 is real.

None of those numbers have changed. What changed is that investors are now willing to wait six to twelve months for that demand to show up in actual company earnings, and they’re selling the rally in the meantime.

Why I’m not closing the trade.

The order books haven’t moved. Rheinmetall just reaffirmed it expects sales to grow 40 to 45 percent in 2026, with profits per share growing roughly 30 percent. When real growth like that arrives, stock prices eventually catch up — sharply, once the consensus turns.

The shares now trade at 18 times next year’s expected earnings, down from 28 times at the February peak. The reset has already happened. We’re paying a much more reasonable price for the same growth story.

What I am doing differently is changing how I get exposure. Instead of buying the broad EUAD basket — which moves up and down with every ceasefire headline — I’m focusing on companies that sell sensors and electronics rather than tanks and missiles.

Those names benefit regardless of which conflict gets prioritized, because they sell capability (radar systems, electronic warfare) that any modern military needs. The specific tickers, the price levels I’m targeting, and the option structures I’m using on Rheinmetall — that’s behind the wall.

Revised parameters.

Confidence: 55% (down from 70%). Asymmetry — the ratio of potential gain to potential loss — now 2.5 to 1 (up from 2 to 1, because the price has fallen while the thesis hasn’t broken).

What would force me to close the trade entirely: a signed and durable Russia-Ukraine framework before June 30 plus a German coalition crisis that re-opens the debate over Germany’s strict borrowing limits. Both have to happen. Neither has, yet.

Verdict.

I was wrong on the timing. I’m not yet wrong on the thesis. That difference is what separates a bad trade from a bad analyst — and which one you are is something readers should be able to judge for themselves.

The three catalysts I flagged on May 4 — now active

Warsh is in.

The 54-45 vote was the tightest Fed Chair confirmation in modern history. The political drama around it (a Department of Justice investigation into the Fed that was later dropped) signals a Fed Chair who’ll have less political space to maneuver than markets initially priced.

His first major appearance is the June meeting of the Federal Open Market Committee — the eight-person panel that sets US interest rates. As of today, the market is no longer expecting any rate cuts in 2026, and some traders are pricing in a non-trivial chance of a hike by December. That’s a complete reversal from where expectations were when I wrote Issue #1, and it explains why my yen prediction (#3) failed and my dollar prediction (#7) is delayed.

Iran is not over.

Ayatollah Khamenei was killed in the late-February strikes. The Strait of Hormuz — the narrow waterway through which roughly 20% of the world’s oil supply moves — remains contested. Brent crude (the global oil benchmark) crossed $104 a barrel on May 11 after Netanyahu’s “not over” warning.

The International Energy Agency has called this the largest oil supply disruption in market history. Until the Strait fully reopens — and analyst estimates of when have slipped from “end of May” to “no clear date” — energy prices stay elevated, inflation stays uncomfortable, and the Fed stays on hold. Which is the engine behind most of what’s wrong in my Issue #1.

Hyperscaler earnings split the market.

(”Hyperscaler” is the industry term for the four biggest cloud-computing companies: Microsoft, Amazon, Alphabet/Google, and Meta.)

Their combined 2026 spending on AI infrastructure came in at roughly $700 billion — higher than I’d projected. AI-related revenue is also scaling: Microsoft’s Azure cloud grew 40% year-over-year, Google Cloud 63%, Amazon Web Services 28%.

Meta dropped 6% on its earnings print, Microsoft and Amazon slipped slightly, but Alphabet rallied on cloud strength. The single-day collapse I was watching for in prediction #1 didn’t arrive. The thesis now pushes into the second half of 2026, depending on what the next earnings cycle shows.

The Decoder — terms I’ve used in this issue

A short reference, in case any of these felt unfamiliar.

The Fed. Short for the Federal Reserve, the central bank of the United States. Sets interest rates, which determines how expensive it is for everyone (businesses, banks, consumers) to borrow money. The Chair is the most powerful unelected economic position in the world.

FOMC. Federal Open Market Committee. The eight-person panel inside the Fed that actually votes on interest rate decisions. Meets eight times a year.

Hyperscaler. The four biggest cloud computing companies — Microsoft, Amazon, Alphabet (Google’s parent), and Meta. Called “hyperscaler” because the scale of their data centers is unprecedented in business history.

DXY. The US Dollar Index. A measure of the dollar’s strength against a basket of six major currencies (mostly euro, yen, pound). When DXY goes up, the dollar is strong. When it goes down, the dollar is weak.

EUAD. The ticker symbol for the Select STOXX Europe Aerospace & Defense ETF — a fund that owns shares of the major European defense companies (Rheinmetall, BAE, Leonardo, Thales, Airbus, Saab, Hensoldt, Dassault). Buying EUAD gives you exposure to the whole sector with one trade.

Asymmetry (in this newsletter). The ratio of what you could gain if you’re right to what you could lose if you’re wrong. An “asymmetry of 5 to 1” means a winning trade pays five times what a losing trade costs. The higher the asymmetry, the less often you need to be right to make money.

Confidence (in this newsletter). My subjective estimate of the probability the prediction lands within its time window. A 70% confidence means I think there’s a 70% chance it works.

Mid-market / private credit. Loans made to mid-sized companies (think $50M to $1B in revenue) by specialized lenders, outside of the public banking system. The market has grown to over $2 trillion in assets and is largely untested in a real downturn.

What’s behind the wall this issue

─────────────────────────────────────────────────────────────
  PREMIUM ACCESS — Issue #1.5 deep-dive
─────────────────────────────────────────────────────────────

  → The 3 new "Warsh-era" predictions I'm adding for the
    back half of Q2 — including one bet on US bond yields
    rising sharply and one pair of sector trades with a
    potential 5-to-1 reward-to-risk profile

  → Full revised analysis on each of the 11 active Q2
    predictions — what's updating, what's holding,
    what's getting closed

  → The exact stocks I'm rotating into for the defense
    thesis: 4 names focused on sensors and electronic
    warfare, plus the specific option structure I'm using
    on Rheinmetall, and the hedge against further EUAD weakness

  → The Hormuz playbook: how I'm positioning for both
    scenarios — Strait reopens (oil could drop 25% in
    ten days) versus prolonged closure (which sectors win)

  → Entry levels, stop levels (where I'd exit if wrong),
    and the conditions that would force me to flip each thesis

  → A live verdict tracker (Notion link) — updated weekly
    with each prediction's status, not just quarterly

  → Bi-weekly mid-quarter updates between now and June 30
─────────────────────────────────────────────────────────────

The honest math, written out.

A 12-month subscription is $X. If even one Tier 1 prediction lands inside its reward-to-risk range — and you express it with, say, $5,000 — the gain at a 5-to-1 ratio is $25,000.

You’d need to be right one time in four to recoup the subscription cost roughly five times over.

That’s the calculation I’d run on any newsletter before subscribing. I’m running it on mine in public.

What you won’t get.

I won’t tell you what to buy.

I won’t tell you how much to put on the table.

I won’t tell you when to take profits.

What you’ll get is the working hypothesis, the instruments I’m watching, the price levels that matter, and the conditions under which I’d change my mind.

Translating that into your own portfolio is your work.

That’s the deal — and it’s what makes this newsletter serious rather than promotional…

Read more

Here’s what makes this different from every other “outlook” you’ve read on the internet: I score myself.

Every prediction from the previous quarter gets a transparent verdict — right, wrong, or still in play.

Since this is issue #1, there’s no scorecard yet. But starting in Q3, you’ll see exactly how this one aged. That’s the whole point — writing under the pressure of knowing we’ll reopen this document in 90 days is what forces honesty.

Before we dive in, one thing I want to make crystal clear:

These are predictions, not investments.

What you’re about to read is my working hypothesis on where I think capital is mispriced and how the next 90 days could unfold. They are starting points for deeper research — not positions I’ve already taken or am recommending you take.

Each of these will need a much deeper dive before becoming a real investment thesis: position sizing, entry zones, hedging, correlation mapping, scenario stress tests, liquidity analysis. That’s the work that comes afterthis document, and it’s what I’ll cover for premium subscribers in dedicated deep-dives over the coming weeks.

Think of this issue as the board on which I’ll be playing — not the moves themselves.

Okay, let’s get into it.

The setup for Q2 2026 is unusually rich.

Hyperscaler AI capex is set to cross $715B this year.

The dollar just broke under 97 for the first time in four years. Copper and uranium are flashing structural deficit signals at the same time.

Europe is rearming at a pace we haven’t seen since the Cold War. And the Bank of Japan finally hiked to 0.75% — the highest level in three decades.

That’s a lot of plates spinning. And historically, when you have multiple regime shifts happening simultaneously, you get extreme dispersion — which is exactly the environment where bold, well-defined predictions either land hard or fail hard. Both are useful.

I’ve ranked the 12 predictions below by risk/reward, from most aggressive to most defensive.

The top of the list contains my highest-conviction asymmetric bets — high potential payoff, but with real probability of being wrong.

The bottom of the list contains ideas I’m more confident in but where the upside is more measured.

That way, you can calibrate based on your own risk tolerance and time horizon.

TIER 1 — High Risk, High Reward

The asymmetric bets. Lower hit rate, but if I’m right the payoff is significant. These are the ones I’ll be watching most carefully.

1. A violent correction (-25% to -40%) on AI hyperscalers during Q2

Let me put some numbers on the table first. Combined capex for Microsoft, Google, Amazon, Meta, and Oracle is projected at roughly $715B in 2026 — that’s about 90% of their combined operating cash flow, up from 65% just last year. To finance this, Morgan Stanley estimates hyperscaler debt issuance will top $400B this year, more than double 2025’s $165B.

Here’s the thing nobody wants to say out loud: capex is now growing roughly 46% faster than AI-related revenue. That gap is wider than the 32% divergence we saw during the 2001 telecom bust. The AI-linked sectors trade at EV/EBITDA multiples near 25x — close to dot-com peak territory. And NVIDIA derives 85% of its revenue from just six customers, with the top four accounting for nearly 60%. That’s binary risk hiding in plain sight.

My thesis: a single hyperscaler cutting 2027 capex guidance — or two consecutive quarters of enterprise AI spending decline — triggers the cascade. The market is pricing perfection. Anything less than perfection is the catalyst.

What would make me wrong: a measurable acceleration in enterprise AI revenue (north of $100B annualized), or four out of five hyperscalers raising 2027 capex guidance in lockstep.

Confidence: 35–45%. Asymmetry: roughly 5:1 if timed correctly.

2. Uranium re-prices structurally toward $130–150/lb

Uranium has already pushed past $100/lb on spot in January, but here’s where it gets interesting: long-term utility contracts are still lagging at $86/lb. Sprott estimates global mine supply will cover less than 75% of future reactor demand. US uranium production fell 44% in Q3 2025. And roughly 70% of post-2027 demand remains uncontracted — the highest level recorded in three decades.

Meanwhile, hyperscalers are signing nuclear PPAs (Microsoft–Constellation, Three Mile Island restart). This isn’t speculative anymore — it’s contracted demand showing up in the data.

The market is still treating uranium as a cyclical commodity. But the structural dependency of AI data centers on carbon-free baseload (90%+ capacity factor, no other source comes close) creates inelastic, multi-decade demand that legacy supply/demand models simply don’t capture.

What would make me wrong: an accelerated restart of Cigar Lake or Inkai combined with a Fukushima-style accident that kills sentiment.

Confidence: 60%. Asymmetry: roughly 3:1.

3. The yen rallies to USD/JPY 140 within 90 days

The BoJ just hiked to 0.75% — the highest since 1995 — and effectively buried Yield Curve Control. Morgan Stanley estimates roughly $2 trillion in yen carry positions are still open globally. As the Fed–BoJ rate gap compresses (Fed median dot at 3.375% by year-end vs BoJ likely 1.25%+), the carry math gets thin.

Here’s what people miss: carry trades don’t unwind because the absolute spread becomes negative. They unwind because path risk becomes intolerable. Volatility on the funding leg goes up, hedging costs go up, and once one big book starts unwinding, everyone else has to follow before the door closes. The August 2024 episode (when BTC dropped 24% in 48 hours on a surprise BoJ move) is the template.

What would make me wrong: BoJ pauses after a weak GDP print, while the Fed holds firm. That keeps the trade alive.

Confidence: 55%. Asymmetry: roughly 4:1 via options structures.

4. A private credit crisis breaks out in mid-market BDCs

Private credit has exploded to over $2T in AUM. Public BDCs trade at NAV discounts that imply real stress, but here’s the part most analysts get wrong: the discount isn’t because of stress, it’s anticipating it. The portfolio composition hides cumulative defaults that monthly NAV markings on private vehicles are smoothing over. The Fed is staying high. The 2026 leveraged loan refinancing wall is real. And the first cliff event is going to be a non-traded BDC gating redemptions.

Once one major non-traded BDC gates, the public BDCs get marked aggressively, and the whole sector reprices.

What would make me wrong: three or four back-to-back Fed cuts that decompress refinancing costs, plus stable EBITDA at LBO-backed portfolio companies.

Confidence: 40%. Asymmetry: roughly 6:1 on a basket short.

Free tier ends here.

  What's behind the wall:

  → Predictions #5, #6, #7, #8, #9, #10, #11, #12 (Tier 2 - Strong Reward, Tier 3 — defensive plays)
  → Specific instruments for each prediction (tickers, options
    structures, futures contracts, ETFs)
  → Entry zones, stop levels, position sizing framework
  → Detailed invalidation scenarios with hard numbers
  → Cross-cutting catalysts to watch (Fed transition, Iran,
    hyperscaler earnings)
  → Bi-weekly mid-quarter updates as theses develop

  → Subscribe to unlock.

TIER 2 — Medium Risk, Strong Reward

Read more