“Cross-sectional dispersion is the only honest signal in late-cycle equity markets. Aggregate index moves are noise; the spread between names that share a label but no longer share a fate is the substance.”

— Lars Toomre, WILT Notebook, April 2026. Attribution confidence: verified. Held in the WKG under brc:fin/cross-sectional-dispersion.

Dateline

Thursday, June 4, 2026, after the close. This edition marks the full pair book to the settled June 4 New York close on exact share counts — no notional approximation anywhere. The book stands at 34 live pairs across eight tranches and gained +$1,210.61 on the day.

Inception-to-date stands at +$290,158.27 — active unrealized +$372,696.27 plus locked realized −$82,538.00 from the May factor-correlation closeouts. The headline +$1,211 is the net of a violently divided tape: the Broadcom short complex (Tranche 7) returned +$48,153 while the memory-and-server tranche (Tranche 6) lost −$34,523. Stripped of the Broadcom short and the SOXS longs, the book lost roughly $47,000 on the day — a single idea carried a bleeding book, and this edition reports both halves.

Section I — Market Dashboard (June 4 Close)

The volatility read is the analytically strange fact of the day: a 12.59 percent single-session break in a megacap with the VIX falling 4.11 percent and VVIX down with it. The market priced Broadcom as idiosyncratic, not systemic — but SKEW rising to 142.15 says tail-risk pricing quietly firmed underneath the compressed headline vol. That divergence is the seed of the BSD assessment in Section VIII.

The United States long end held: the 30-Year Treasury at 4.98 percent with the 10–2 spread near +40 basis points. The non-US comparators carry the ⚠ flag — the May 18, 2026 Global Sovereign Credibility Event drove the UK 30-Year Gilt to its highest since 1998, the German 30-Year Bund to multi-year highs, and the Japan 30-Year JGB to an all-time-high regime; the June 4 daily closes for those require Tier-1 confirmation and are shown for context only.

The breadth inverted. The Dow closed at a new all-time high and the Russell 2000 rose 1.45 percent — the best small-cap session in three weeks — while the Nasdaq Composite finished flat and the Philadelphia Semiconductor Index fell. This is the mirror image of the late-May narrowing thesis: breadth broadened even as the largest mega-cap names struggled. The macro counterpoint to the Broadcom break is that the selloff was contained to a handful of target-raise names while the median stock rose.

Section II — Market Movers (Close)

The single name to watch is NVIDIA. It reversed to green (+1.82%) on a day the rest of the complex fell, and that is the strongest available evidence for the bottleneck-migration thesis: had NVIDIA broken with Broadcom and Micron, the read would be a complex-wide de-rating rather than a target-raise-specific one. It did not. The thesis lives or dies on that distinction.

Section III — Standing Watchlist Monitors (Close)

Mag Seven internal dispersion at close: Microsoft (data pending), Alphabet +3.68%, NVIDIA +1.82%, Meta (data pending), with Apple, Amazon, and Tesla pending. The confirmed names span roughly 350 basis points on the session; adjacent name Broadcom −12.59%. The label covered divergent stories; today's dispersion is the widest of the past two weeks.

GSIB and consumer-credit gating: the P8 short leg is a Global Systemically Important Bank basket marked directly at 57.68 / 58.68 (not by XLF proxy). Goldman Sachs and American Express are tracked as the consumer-credit rotation tell; the financials bid into the close is consistent with rotation away from concentration-risk names.

Silver / GROUP-17 bullion complex: Silver eased to $72.57 and Gold firmed to $4,475.80; the Paper-versus-Physical thesis took a one-day unwind but the structural backwardation argument is intact.

Section IV — The AI Cross-Section Differential

The most analytically rich feature of the day is invisible in any index move. Within the AI-infrastructure complex, five names moved as follows: Broadcom −12.59%, Micron −7.74%, AMD −3.56%, NVIDIA +1.82%, Dell +0.23%.

The test, defined. A name is “cash-flow-visible” when its current guidance already discloses the demand the market is paying for — Dell after its May 29 disclosure of a $51.3 billion AI backlog, NVIDIA on near-term shipment cash flows. A name is “target-raise-momentum” when its multiple prices an expected future guidance raise not yet delivered — Broadcom on the standing $100 billion fiscal-2027 AI target, Micron after the UBS target revision to $1,625. On June 4 the first category held and the second broke. That is the bottleneck-migration thesis: cash flows reprice in sequence as they work through the stack, and the names still pricing the next raise are the fragile ones.

Why Tranche 6 lost −$34,523. This is not a repeat of the May factor-correlation failure that forced the P7/P20/P22 closeouts. There the long and short legs were the same factor (both long-AI-infrastructure), so they fell together. Here the construction held: P28 (Micron long / Dell short) is an intra-factor spread, and it lost precisely because the cross-section differentiated as designed — the long was the target-raise name and the short was the cash-flow-visible name. The spread worked the wrong way, but it worked as a spread. The loss is a directional miss, not a correlation failure; the distinction matters for whether the construction is sound.

Section V — What Drove the Tape

Broadcom led the break. AVGO settled −12.59% at $418.91, recovering from a morning print near −15 percent; organized buyer interest emerged below $410 even as the structural narrative held. The pair book's P36 and P37 short legs captured −12.59 percent on a combined 626 AVGO shares.

The Corning cluster dragged the long book. Corning fell 1.52 percent, and because four separate pairs hold GLW long (P1, P6, P13, P21), that single move produced four separate pair losses totaling roughly −$10,200. This is a factor-concentration observation hiding in plain sight: the book is long Corning four ways, and a quiet down day in one name compounds across the tranche structure.

Financials and small caps caught the rotation bid. Goldman Sachs and American Express rose on the consumer-credit rotation; the Russell 2000 had its best session in three weeks. The short legs in P32 (AXP) and the materials cyclicals worked against the book as the rotation lifted exactly the names several pairs are short.

Section VI — Operation Epic Fury: The Macro Read

The Strait of Hormuz remains closed to commercial transit, Day 97 by the inclusive count from the February 28 closure. WTI settled at $93.19 and Brent at $95.41; WTI front-month settled $93.19 with M1-M3 backwardation of roughly +$2.10/bbl; Brent $95.41, the spread holding spot over deferred. The crude move was consensus pricing a ceasefire-progression read rather than a physical-supply reassertion — jet fuel did not rally with futures, and the vessel-traffic data has not improved.

The Strategic Petroleum Reserve now carries a number: 357.1 million barrels (week ending May 29, 2026; down roughly 50 million barrels (about 12 percent) since the February 28 closure; lowest since April 2024). ⚠ Tier-2; weekly EIA print. Each weekly draw narrows the political-management runway toward the late-third-quarter convergence window the desk has named as its central second-event risk. The three conditions for a durable long-end Treasury rally remain unmet: the strait stays closed, Brent trades above 85 dollars, and core inflation has not printed below 2.5 percent.

Section VII — The Pair Book at the June 4 Close (Exact P&L, 34 Pairs)

The book is marked from exact entry share counts to the June 4 settled close. Day P&L is long shares × (June 4 − June 3) plus short shares × (June 3 − June 4). Each row shows the exact entry (shares @ price), the day move, day P&L, and position-to-date from entry. The six thin-quote legs are marked at their reliable end-of-day settled closes. P8 short is the GSIB basket marked direct (57.68 / 58.68).

Lineage notes: P36 is the reopen of closed P22 (Chevron/Broadcom) under a new identifier — the immutable closed-record of P22 (−$8,239 realized) is preserved separately. P33 short is held as 8,718 International Consolidated Airlines ADRs struck in GBP (426.10p at 1.3460 on June 2 London open); its inception mark carries an FX component and is marked to the USD ADR close intraday. Pair count: the canonical register enumerates 34 active lines; the ledger header states 33. This edition uses the table count of 34 and flags the one-line discrepancy for register reconciliation rather than resolving it silently.

Per-Leg Dollar Attribution — Headline Pairs (Day)

The decomposition isolates which leg carried each pair. P28's −$18,182 loss is almost entirely the Micron long (−$17,717), not the Dell short (−$466) — the target-raise long is the damage. P37's gain is both legs working: the SOXS long rose as semis fell and the Broadcom short paid.

Tranche Day P&L by Vintage

The vintage view is the day's structure in one glance: the newest tranche (T7, initiated this week) drove the entire gain, while the oldest and flagship tranche (T1, intact since September 2025) was the largest detractor on the Corning/Generac long legs. The book made money on its most recent conviction and gave some back on its foundation.

Inception-to-Date Reconciliation

Section VIII — Book Risk Flags

Section IX — Momentum: RSI Snapshot at the June 4 Close

14-day Wilder RSI at the June 4 close. The mechanism behind the target-raise-expectation break is visible here: the names that broke were the overbought ones. Note NVIDIA sitting mid-range — the non-confirmation that anchors the cross-section thesis.

The book carries a deliberate, monitored tension: it is long the overbought AMD through P19 while short the complex through the SOXS legs. P19 is a relative-value long against a Korean short; the SOXS legs are a directional complex short. Not contradictory, but the net book is long the most overbought chip names while shorting the complex — named here, not hidden.

Section X — BSD Second-Event Risk Assessment

⚠ Bull Shit Detection (“BSD”). First event (established): the closed Strait of Hormuz, Brent above 95, a held long end. The June 4 tape sharpened one second-event candidate above the others.

BSD convergence read. June 4 is the cleanest single-day demonstration of the target-raise-expectation mechanism the May 29 deep edition flagged: a beat-and-no-raise produces a non-linear single-name break. But the VIX compression and broadening breadth say the market priced it as idiosyncratic, not systemic. The structural argument is that single-name target-raise reactivity, repeated across the complex over several earnings cycles, eventually reprices systemic volatility — the SKEW tick to 142 is the first faint evidence. Tomorrow's NFP is the next inflection.

Section XI — Vocabulary Corner

Section XII — Book of the Day

Section XIII — Forward Calendar

Section XIV — Standing Governance Notice

Section XV — Author’s Note

This is the corrected, complete June 4 close edition. It supersedes both an earlier navy-styled close edition (wrong pair identities from a laptop/mobile numbering collision, P&L on notional approximation) and the first exact-but-thin final. A separately posted June 4 morning edition argued the tape held two parallel Castle Bravo demonstrations; the close confirms both held in direction with partial absorption in magnitude — Broadcom recovered from −15% to −12.59%, crude from its intraday low.

The thesis stated at the top held through the tape: one label, four outcomes, and the dispersion is the signal. The book gained +$1,210.61 on a single idea — short the AI complex through the Broadcom short and the SOXS longs — while the Corning longs and materials cyclicals bled. Inception-to-date is +$290,158.27, re-struck against the canonical 34-pair register with the realized total corrected to −$82,538. The discipline that matters: an exact figure is never softened, a thin name is marked at its reliable settled close, a numbering error is corrected in the open, and the day's best analytical observation leads the edition rather than hiding in the closing note.

The Coffee Grind remains in the final months of an internal-only run that transitions to a paid daily under the ProvokAI banner in the third quarter. The first ~400 words of every edition stay free; full editions will be subscription-tier (target $200/month, roughly $10 per trading day). Beta-reader feedback to lars@brcfintech.com.

— Lars Toomre, Palm Beach Gardens, Florida · Thursday, June 4, 2026