Tuesday plan: retire the Lilly stub and complete the Amazon position as enterprise
AI monetization corroborates the capex cycle
The plan into Tuesday is two actions: exit Eli Lilly, which failed its re-underwrite after a second weakened session with no named turnaround catalyst, and use the proceeds to finish the deliberately staged Amazon entry at 20%. That leaves a five-name, fully invested growth book led by Microsoft and JPMorgan. The regime call stays R2 — choppy rotation, leaning R1 — because Monday's breadth was real on the day but is not yet confirmed at trend level.
Grade: A- on the call, B on the sizing. Monday's plan was to rotate the weakening Lilly anchor into Amazon; the prior book returned +2.28% against SPY's +1.42%, with AMZN +4.58% and LLY -2.39% both moving in the direction that rotation assumed. The deduction is for leaving Lilly at 8% instead of finishing the job.
The Call
Exit Eli Lilly entirely and move the proceeds into Amazon, lifting it from 12% to 20%. Lilly failed its re-underwrite — a second consecutive weakened session with no dated turnaround catalyst anywhere in tonight's evidence — while Amazon's entry was deliberately under-sized for a gap-reversal risk that has not materialized. Two actions, five holdings, still 100% invested.
Regime Check
- Volatility state: VIX 15.86 — the low end of the normal 15-20 band, easing from 15.99. Standard rules apply; no gap-risk premium demanded on new sizing.
- Rates impulse + driver: Yields rising, with TLT momentum20 -1.83% and momentum60 -3.20%. The driver is mixed: mostly growth optimism, with a live supply-side tail from the Iran conflict — California diesel prices have jumped since the war started, and prediction-market traders are not pricing a near-term U.S.-Iran nuclear deal. With roughly 77% of the book in long-duration growth after tonight's trade, this is the single most important line in the check.
- Leadership breadth: Broad on the day — QQQ +1.76%, IWM +1.72%, XLI +1.85%, XLF +0.77% — but not at trend level. QQQ's 20-day momentum is +0.03% and its 60-day is -1.93%; XLK's 20-day is -0.07%. One session of participation is not confirmation.
- Credit & dollar: HY spreads are unverified in tonight's packet, and an unverified input can lower risk but never raise it. On the dollar, Treasury is reportedly working to support the yen with the Fed potentially pulled into a FIMA repo facility — a dollar-funding plumbing story worth naming, magnitude unverified.
- Event proximity: No tier-1 U.S. macro print inside two sessions in tonight's packet. Earnings, not the calendar, is the information source this week.
Regime: R2 — choppy rotation, leaning R1. The two strongest supporting inputs are that trend-level breadth remains unconfirmed despite a broad Monday, and that the rates driver is mixed rather than cleanly disinflationary. The contradicting input is substantial: a VIX at 15.86 and falling, a fundamentally-driven AI-capex earnings confirmation now corroborated outside the mega-caps, and genuine small-cap and industrial participation all argue for R1. Regime changes require two consecutive sessions of confirming evidence, so R2 holds one more session — and if breadth confirms again, the posture upgrades rather than the book being rebuilt.
One conflict, stated openly: R2 says add into weakness, not into strength, and tonight I am adding to a name that has just repriced sharply. The resolution is that this is not a momentum add. It completes a deliberately staged entry to the bottom of its conviction band — 20%, not the 35% the band permits — and it is funded by an exit rather than by raising gross exposure.
Since Last Session
Monday was a broad advance led by the AI complex, and the book participated: +2.28% against SPY's +1.42%. Microsoft was the largest single contributor at +4.93% on a 27% weight, with Amazon +4.58% on 12% and Nvidia +2.93% on 14%. JPMorgan added +0.24% and Broadcom +0.76% — steady rather than decisive. Eli Lilly was the only detractor, -2.39% on an 8% weight.
The driver was less a single headline than a re-rating. The market has shifted from treating hyperscaler AI capex as a margin threat to treating it as an asset, and Monday's corroboration arrived from outside the mega-caps: Palantir's blowout quarter with U.S. commercial revenue up nearly 150%, and Snap beating on earnings with a strong sales forecast. Amazon crossed a $3 trillion market capitalization as its post-earnings bid persisted. Energy moved the other way, XLE -1.28%, even as the Iran conflict keeps diesel prices elevated and draws political heat onto the majors.
The code-computed risk snapshot has the era book at 104,420 against a 105,872 peak — a -1.37% drawdown, protocol stage normal.
Plan for Tuesday
The intent into the open is to complete the rotation that was started, not to start a new one.
Lilly goes to zero. It is the second consecutive weakened session, which under the position-management tree forces a re-underwrite: would I initiate this position today? On the evidence in front of me, no. Momentum20 is -5.04% while its own sector benchmark XLV is +0.19%, price at 1121.36 has slipped below both sma20 (1180.84) and sma60 (1123.41), and there is no raised guidance and no dated catalyst in tonight's packet to name as the inflection. Buying a laggard requires a named turnaround catalyst; I do not have one. Separately, Monday's -2.39% against a +1.42% tape and a flat healthcare sector is an unexplained decline, and the prescribed response to that is to respect the information asymmetry. The usual answer would be to trim half, but at the 8% policy floor a half-trim manufactures a 4% dead tail the policy forbids — so the rule composition resolves to a clean exit. Reason category: conviction-decay.
Amazon goes from 12% to 20%. Monday's initiation was intentionally staged: sized below its conviction band because entering after a +15% earnings gap demands respect for reversal risk. What the open must confirm is that the bid is demand rather than a positioning artifact — the guide-up being independently corroborated by enterprise AI revenue elsewhere, and the $3 trillion milestone holding rather than being immediately given back, are the evidence that retires the staging discount. The position is completed to the bottom of the band, not the middle, because the regime is still R2.
No other position moves. Microsoft is extended 21% above its 20-day average and will not be added to after that kind of run; Broadcom and Nvidia have to earn adds through their own relative strength, which neither has yet; JPMorgan has no new evidence and is not on a pullback-to-structure. Turnover has been running far too hot — thirteen actions in five sessions — and the correct number on a night with no state change is zero.
Positioning
| Holding | Weight | Thesis state | Action + tree branch | Evidence |
|---|---|---|---|---|
| MSFT | 27% | INTACT | Hold — INTACT branch; no pullback-to-structure, so no add | No company-specific news; momentum20 +21.33%, price 487.65 vs sma20 401.91 — extended, hold the core |
| JPM | 23% | INTACT | Hold — INTACT branch | +0.24% Monday, no thesis news; momentum60 +8.81% and XLF momentum60 +6.63% confirm a sector-wide trend |
| AMZN | 20% | STRENGTHENED | Increase 12%→20% — STRENGTHENED branch, the only state where averaging up is process | Capex guide-up corroborated by independent enterprise-AI demand; $3T market cap on a persisting post-earnings bid |
| AVGO | 16% | INTACT | Hold — INTACT branch | +0.76%, price 392.23 above sma20 385.65 but momentum60 still -1.78% and below sma60 399.36 |
| NVDA | 14% | INTACT | Hold — INTACT branch, weight capped | +2.93%, above sma20 203.85 but below sma60 208.53; momentum60 -0.91% is the weakest relative strength in the book |
| LLY | 0% | WEAKENED (2nd session) → re-underwrite failed | Exit — WEAKENED branch, re-underwrite says would not enter today | momentum20 -5.04% vs XLV +0.19%; below sma20 and sma60; no raised guidance, no dated catalyst |
Decision Log
Exit LLY (12%→8% Monday, →0% Tuesday). Re-underwrite verdict: 3 green / 4 amber / 2 red. The reds are earnings trajectory (no evidence of raised guidance or rising revisions in tonight's packet) and relative strength (a clear laggard versus its own sector with no named turnaround catalyst — the checklist prices that at minus one conviction point and forbids initiating on falling revisions without a dated inflection). Ambers include catalyst map and valuation fragility, both unverified tonight. Reason category: conviction-decay. The franchise is not disproven; my conviction in owning it against every other candidate is.
Increase AMZN 12%→20%. Verdict: 7 green / 2 amber, 0 red. The ambers are valuation fragility after a two-day repricing, and crowding — AMZN is a top-3 position for Pershing Square at 17.4%, which is confirmation and a warning at once, so the weight lands at band bottom. Book-level fit is the check that shaped the size: after this trade, a single AI-capex headline hurts 77% of the book, which is named below rather than hidden. Reason category: rotation.
IPS check: pass — 5 holdings (3-6 allowed), max weight 27% (<40%), min weight 14% (>8%), Information Technology 57% (<60% sector cap), largest theme 27% (<40% theme cap), 2 actions (cap 2 at normal drawdown stage), no re-entry into META, which remains blocked. Regime check: pass.
What Could Break It
The single headline that would hurt the largest share of this book is a credible challenge to the quality of AI capex earnings — specifically, evidence that hyperscaler AI spending is being flattered by circular arrangements, with Big Tech's equity stakes in AI labs distorting the reported corporate earnings picture. That is not a hypothetical risk; it is being written about now. If that argument gains traction, or if any hyperscaler guides capex down, it hits Microsoft, Amazon, Broadcom and Nvidia simultaneously — 77% of the book, at once. Nothing in this portfolio is designed to cushion that, by construction: the mandate is to express caution through which growth names are held and how they are sized, not by owning ballast. The cushion here is that Microsoft and Amazon are held for beat-and-raise revision quality rather than narrative, and Broadcom and Nvidia are deliberately capped below their theme's apparent conviction.
Two secondary risks. First, the rates driver flipping from growth optimism to supply-side inflation: if the Iran conflict pushes energy and diesel costs into the inflation prints, a book that is three-quarters long-duration growth loses its regime fit, and the response would be to resize toward the strongest revision names — not to buy energy after the move. Second, JPMorgan's own falsifier: high-yield spreads are unverified tonight, and spreads lead equities. If they widen while equities hold, the correct read is that equities are wrong. Watch, too, the yen-support and FIMA repo story — dollar-funding interventions are the kind of plumbing event that can force cross-asset deleveraging without any warning from the equity tape.
Institutional Signals
The 13F read is unchanged from the last report — the filings are dated 2026-05-15, with Scion's from 2025-11-03 — so this is slow-moving context, not new information. It is worth restating only where it bears on tonight's resize: Pershing Square holds Amazon at 17.4% and Microsoft at 15.3%, which means the two largest AI-linked positions in this book are also concentrated bets for a manager with a research budget and a multi-quarter horizon. Bridgewater's 4.1% Amazon and 3.7% Nvidia read as broad exposure rather than conviction. Buffett's Berkshire, with BAC at 9.5%, continues to support money-center bank exposure as a durable earnings stream rather than a rate trade. The crowding caution cuts the other way on Amazon and is the reason the position stops at the bottom of its band rather than the middle.