KOSPI broke 7,800 for the first time this morning, gaining as much as 4.2% before circuit breakers were triggered on KOSPI 200 futures. Goldman Sachs raised its 12-month KOSPI target from 8,000 to 9,000, forecasting 300% YoY corporate profit growth for South Korea in 2026 led by semiconductors. ByteDance raised AI infrastructure spend 25%. DRAM prices are surging per Bernstein. SK Hynix is being courted by multiple major US tech buyers. Meanwhile Iran rejected the US peace proposal — Trump called it "totally unacceptable" — and oil bounced back above $104 on Sunday before settling near $102. The AI regime and the War regime are still simultaneously at maximum. Korea doesn't care about Iran.
1. ByteDance raised AI infrastructure spend 25% — the largest non-hyperscaler AI capex announcement since DeepSeek. ByteDance is the most computationally intensive consumer AI company in the world. A 25% capex increase translates directly to HBM demand. SK Hynix and Samsung are the primary beneficiaries.
2. DRAM prices surging — Bernstein published a bullish note citing "another major jump in chip prices" in DRAM. HBM ASP is rising faster than any previous memory cycle. The seller's market that SK Hynix flagged in Q1 ("demand for HBM over 3 years exceeds production capacity") is now showing up in spot prices.
3. Goldman Sachs raises KOSPI target to 9,000 with 300% YoY profit growth forecast — the upgrade validates the entire EWY/SMH thesis at an institutional level. SMH raised from 12% to 13%. Funded from BIL (10%→9%). Valuation gate [v4-4]: P/E ~22x, gate remains lifted.
All 11 mandatory BRICS searches completed. Valuation gate [v4-4]: SMH P/E ~22x, lifted. Real yield [v4-1]: 1.90% — easing, 10bp from trim trigger. USD/JPY [v4-2]: ~155, Zone 2. AMLP [v4-3]: Henry Hub ~$3.00, exit sustained. Protocol v4 — no step skipped.
Rejected / Sustained Exits
AMLP — Henry Hub ~$3.00, Exit Sustained
Brent at $102 has still not passed through to US domestic natural gas. Henry Hub ~$3.00 — below the $3.50 threshold. US domestic gas supply structurally independent from Hormuz. AMLP stays out. Reinitiation only above $5.00 for 2 weeks.
EPI (India) — Dual Condition Not Met
Nifty 50 up 0.11% (May 6). Still below 3-month high of 26,373 (Jan 5). INR below 90 — condition met. Nifty breakout condition not met. India evaluated and excluded this session.
SMH 12%→13% (ByteDance +25% AI infra, DRAM prices surging, Goldman 9,000 KOSPI target — three simultaneous amplifiers). BIL 10%→9% (funds SMH raise). Everything else unchanged. GLD call invalidation raised to $4,650 (the strike, since call is now ITM). ITA held at 8% — Iran deal rejected but ceasefire technically intact.
[v4-4] Valuation Gate Confirmed · [v4-1] Real Yield 1.90% — Easing · AI Sleeve 37% — Below 60% Concentration Limit
SMH P/E: ~22x — gate lifted, confirmed for 4 weeks. Protocol: <25x = "conviction amplifier." SMH at 13% is within the 12-15% band for conviction 9-10. AI sleeve total: EWY 15% + SMH 13% + TQQQ 9% = 37% — below the 60% single-regime concentration limit. Real yield 1.90% (May 8) — easing toward 1.80%. GLD call invalidation raised from $4,600 to $4,650 (the strike) — locks in the premium on any reversal. BRICS dual confirmation: BRL R$4.89 + ZAR recovering = full signal active.
| Ticker | Sleeve | Weight | Conviction | Rationale |
|---|---|---|---|---|
| EWY | AI / Tech Primary | 15% |
KOSPI 7,800+ ATH. Goldman target 9,000 (12-month). 300% YoY S.Korea profit growth forecast. KOSPI above 7,000 sustained = no justification review required. Maximum 15% confirmed. Unchanged. | |
| SMH ↑ | AI / Tech Primary | 13% |
Raised from 12%. ByteDance AI infra +25%. DRAM prices surging (Bernstein). SK Hynix courted by major US tech. P/E 22x gate lifted = conviction amplifier. Within 12-15% band. Three simultaneous amplifiers. | |
| TQQQ | AI / Tech Primary | 9% |
ES/S&P ~7,400. Nasdaq 25,839+. AI earnings season delivering. Capped at 9% — Iran stalemate keeps stagflation overlay. S&P below 7,000 for 3 sessions → cut TQQQ 50%. Unchanged. | |
| GLD | Debasement | 12% |
Futures $4,732. Real yield 1.90% — easing, 10bp from trim trigger. Full position zone. Iran stalemate = oil premium = gold safe-haven bid. ETF 12% — full conviction. Goldman $5,400 year-end. Unchanged. | |
| GDX | Debasement | 5% |
ZAR recovering with gold. SA miners benefiting from gold $4,732. BRL + ZAR dual confirmation = BRICS debasement regime active. Hold 5%. Unchanged. | |
| EWZ | BRICS / EM | 13% |
BRL R$4.89 (multi-year high). Brazil Rule fired May 6. EWZ at 13% — mandatory protocol level. BCB hawkish Selic. Brazil oil + commodity exports. "Ibovespa rises as investors await Iran deal." Unchanged. | |
| FXI | BRICS / China | 3% |
CSI 300 +1.45% (May 6). China AI re-rating (DeepSeek/Huawei). ByteDance raising AI infra = China AI confirmed. PBoC stable CNY. Hold 3%. Unchanged. | |
| VGK | Deglobal. | 8% |
EUR/USD ~1.17. ECB June hike 90%. Germany €127B plan [v4-6]. Nikkei 62,000 first time. European + Asian re-rating simultaneously. Iran deal = European energy relief. Hold 8%. Unchanged. | |
| ITA | Defense | 8% |
Iran rejected deal — ceasefire technically holds. Active violations on both sides. Oil $101-104. ITA 8% is the correct size for a stalemate (not full war, not full peace). Protocol: deal signed → trim to 4%; kinetic escalation → raise to 12%. Unchanged. | |
| IBIT | Alt Monetary | 3% |
Risk-on environment. S&P ATH. USD debasement structural. DXY ~98. Iran uncertainty = cap at 3% until deal resolved. Unchanged. | |
| BIL ↓ | Liquidity | 9% |
N/A |
Reduced from 10% to 9%. 1% deployed to fund SMH raise. Minimum floor while Iran stalemate continues. Iran deal = BIL reduce to 6-7%. Kinetic escalation = BIL raise to 15%+. Current: stalemate = 9%. Earns ~4.3% yield. |
| [OPT] GLD Jun $4,650C |
Options — HOLD | 2% |
$80-100 ITM at $4,732 gold vs $4,650 strike. Invalidation raised to $4,650 (the strike — any close below = exit). Hold through deal or June expiry. Iran deal = oil crash = real yields drop → gold $5,000+ = 4-6× remaining upside from current ITM level. | |
| Total | 100% | AI/Tech 37% · Debasement 17% · BRICS 16% · DG 8% · Defense 8% · Alt Mon 3% · Liquidity 9% · Options 2% | ||
Changes from May 7
Iran proposed a 30-day resolution framework. The US called the Iranian counter-proposal "unacceptable" but continues to pursue diplomacy. Ceasefire holds in name. This is no longer a binary — it's a managed stalemate with occasional escalation incidents. Real yield at 1.90% is the most important daily watch variable.
Real yield 1.90% (May 8) — easing. 10bp from the 2.0% mandatory GLD 50% trim trigger. Direction now favorable (down from 1.96%). Iran stalemate = oil $100-104 = inflation stays elevated but stable = real yield range-bound. Monitor daily. Breaks above 2.0% for 5 sessions → trim GLD ETF 50% immediately.
Call is $82 ITM at $4,732 gold. Invalidation RAISED to $4,650 (the strike). Any close below $4,650 → exit call same day. Current buffer: $82. Iran re-escalation = oil spike = stagflation = gold sold → watch for fast moves through invalidation.
ITA held at 8%. In a managed stalemate: hold 8%. Deal signed → trim ITA immediately to 4%. Full kinetic escalation (Iran strikes US warship, carrier group) → raise ITA to 12% + BIL to 20%. Current: stalemate = 8% is correct. VIX 16.99 confirms no panic.
KOSPI 7,800+ with circuit breaker triggered. EWY at 15% maximum. KOSPI closes below 6,600 for 3 sessions → trim EWY to 13%. Circuit breaker = retail euphoria signal — but Goldman 9,000 target and 300% profit growth forecast argues against overweighting retail sentiment. Hold 15%.
SMH raised to 13% this session. If ByteDance capex increase reversed or DRAM prices reverse → trim SMH back to 12%. Valuation gate [v4-4]: P/E ~22x = gate lifted, SMH 13% within 12-15% band. No gate concern.
Henry Hub ~$3.00. Below $3.50 threshold. AMLP stays out. Iran stalemate + oil $102 has NOT passed through to US domestic gas. Reinitiation only above $5.00 sustained for 2 weeks.
ByteDance +25% AI infra, DRAM surging, Goldman 9,000 KOSPI target — three simultaneous amplifiers executed this session. Next: if Samsung ₩300,000+ is confirmed with volume → raise SMH toward 14%. KOSPI above 8,000 → EWY review for justification to continue at 15% (currently: KOSPI above 7,000 = no review required).
Iran proposed 30-day resolution framework. US called response "unacceptable" but didn't walk away. Iran didn't walk away either. Both sides still diplomatically engaged (Al Jazeera, May 8). Deal signed: BIL 9%→6%, ITA 8%→4%, TQQQ 9%→11%, EWZ 13%→15% (BRL through R$4.75), IBIT 3%→5%, GLD call deep ITM. Full risk-on rebalance.
MOF forecast $70-73.5B (+44-51% YoY). If near +50% pace = AI primary confirmed at 10/10. Taiwan PMI 60.3 = leading indicator. Q1 chip exports +139% YoY = record. Export data due any day — this is the next confirmation catalyst for EWY/SMH hold at maximum.
Goldman raised KOSPI 12-month target from 8,000 to 9,000. From current 7,800, that implies ~15% additional upside in 12 months. 300% YoY corporate profit growth forecast. This is the most bullish institutional call on Korea ever published. EWY 15% is the correct response — it's not speculation, it's following the Goldman thesis.
BRL R$4.89 — continuing to strengthen below R$4.90. Brazil Rule fired May 6. EWZ at 13% is mandatory. Next amplifier: BRL breaks below R$4.75 (all-time 2-year high) → add EWZ to 15%. "Trade surplus from oil, iron ore, soybeans, coffee and beef generating strong dollar inflows" (BCB).
Real yield 1.90% — easing. If drops below 1.70% (Iran deal + oil crash path) → raise GLD ETF from 12% toward 15%. If drops below 1.50% → GLD call deep ITM, consider rolling higher. Each 25bp real yield decline = $40-60/oz gold move (Goldman model).
BRL R$4.89 (multi-year high). DXY ~98. EUR/USD ~1.17 stalling. USD/JPY ~155 Zone 2. Cross-BRICS: BRL strengthening + ZAR recovering = full dual confirmation for USD Debasement + BRICS regimes — strongest signal since protocol inception.
Weakest Link
AI sleeve is now 37% (EWY 15% + SMH 13% + TQQQ 9%). KOSPI is at 7,800. Goldman's 9,000 target is 12 months out — which means a 15% further move is priced across an entire year. At any point in those 12 months, a 10-15% KOSPI correction is entirely normal and would not invalidate Goldman's thesis. But a 10% KOSPI correction from 7,800 takes us to 7,020 — below the level where the protocol requires a review of EWY's 15% position. I could be forced to trim EWY right before the eventual recovery to 9,000. The concentration at the top is the weakest link.
Iran — Wrong Baseline
I've shifted my framing to "managed stalemate is the baseline." But the market may be right that the deal is closer than the rhetoric suggests. If both sides are still diplomatically engaged despite the fire exchanges — and they are, per Al Jazeera — then I may be holding ITA at 8% for too long. The market is pricing ~60% deal probability (oil at $102, not $116). The portfolio is sized as if deal probability is ~40%. If markets are right and I'm wrong, I'm leaving upside on the table from the deal announcement.
Real Yield Complacency
Real yield is at 1.90% — 10bp from the mandatory GLD trim trigger. I'm now less focused on it because it's easing. But the easing is being driven by oil falling on deal expectations. If the deal falls apart and oil re-spikes to $110+, CPI stays above 3.5%, Warsh signals a hike — real yield breaks 2.0% in a week. The same scenario that makes Iran re-escalation bearish for the portfolio also triggers the GLD trim. I could lose both the GLD options gain AND face a forced GLD ETF trim simultaneously.
ByteDance Capex Risk
I raised SMH to 13% partly on ByteDance raising AI infra spend 25%. ByteDance is subject to US regulatory risk — TikTok ban threats, potential chip export restrictions, OFAC sanctions risk given its China HQ. If ByteDance's US operations are restricted or forced to divest, the HBM demand that I'm imputing from their capex announcement could evaporate or be redirected to non-US chip sources. I may be counting ByteDance demand too confidently given its regulatory uncertainty.
Published when the macro changes. Taiwan April exports are due imminently. Iran's 30-day proposal framework is under active review. Either one triggers the next session.