The Sovereignty Cap (E10) — fab + geopolitical risk.

Most insurance is for events you've seen happen — house fires, car accidents, theft. The historical base rate guides the premium. The Sovereignty Cap (E10) is insurance for a different kind of event: one with a base rate of zero in modern markets but a magnitude that would make every other risk pillar irrelevant if it happens. Specifically: a kinetic event in the Taiwan Strait disrupting TSMC fab capacity. Most retail traders ignore this because it's never happened in the era they've been trading. The framework treats it as the one geopolitical risk that's both large enough and concentrated enough in specific names to warrant a separate sleeve cap on top of all the other gates.

Why this specific risk gets its own pillar

Most geopolitical risks (sanctions, regional conflicts, trade tensions) are diffuse — they affect markets broadly but don't concentrate in identifiable single-name exposures the way Taiwan does. The Taiwan/TSMC concentration is unique: a handful of US-listed names depend on TSMC's continued fab operation for a material fraction of their revenue or technology pipeline. NVDA's leading-edge GPU dies are fabbed there. AVGO's most advanced chips. AAPL's M-series silicon. AMD's Ryzen and Instinct lines. TSMC itself trades as a US ADR. ASML's EUV business depends on TSMC as a primary customer.

If Taiwan goes kinetic — anywhere from blockade through full-scale military action — every single one of those names experiences a step-function repricing that has nothing to do with the company's fundamentals or the trader's R:R math. The pre-flight chain's R:R floor doesn't help; the gap from such an event would be 30-60% in single-day moves on the affected names, well past anything the per-trade math is sized for.

The Sovereignty Cap mechanic

The framework treats Taiwan-fab-exposed names as a single sleeve regardless of how they classify under GICS sector or sleeve-allocation rules. The cap:

Sovereignty Cap (E10)
  Names tagged: NVDA, TSMC (TSM), AVGO, AAPL (partial weight), AMD,
                ASML, MU, partial weighting on others
  Cap: 25-30% of total portfolio risk-weighted exposure
  Composes WITH the per-sector cap (L14) and per-sleeve cap (L13) —
    all three apply simultaneously

If you have NVDA (15%), AVGO (10%), TSM (5%), and a starter in AMD (3%) — that's 33% of the portfolio in Taiwan-fab exposure even though sector classification spreads them across multiple GICS buckets. The Sovereignty Cap fires; new entries on any name in the tagged set get refused at preview.

The kinetic-risk model — what it actually estimates

The framework doesn't try to predict a Taiwan event. It models conditional impact given an event. The math:

• For each tagged name, estimate the % of revenue / EPS contribution dependent on TSMC fab capacity
• Apply a step-function shock — assume 6-18 months of partial fab disruption, with a 40-60% revenue impact during the disruption window
• Translate to share-price impact via consensus 12-month forward earnings multiples
• Sum across the portfolio's tagged exposure to get a worst-plausible portfolio shock

The output isn't "this is going to happen." The output is "if this happens, your tagged exposure produces an X% portfolio loss in a single repricing event." The 25-30% Sovereignty Cap is calibrated so that the modeled worst-case shock stays under 12% portfolio impact — survivable, not catastrophic.

What the cap doesn't do

The Sovereignty Cap is not a directional bet. It doesn't say "don't own NVDA" or "Taiwan will go kinetic." It says: "don't have so much of the portfolio in Taiwan-fab-dependent names that a single low-probability-but-high-magnitude event is portfolio-ending." If your portfolio is 12% NVDA and nothing else tagged, the cap doesn't fire. If you want to add AVGO at 8% on top of that, the cap considers whether 20% combined exposure is acceptable — usually yes, well below the 28% effective cap. If you then want to add AMD at 5% and TSM at 6% — that's 31% combined; the cap fires.

Why 25-30% specifically

Above 30% tagged exposure, the modeled worst-case shock crosses 13% portfolio loss in a single event — past the framework's drawdown defense layer (Lesson 15). Below 20%, the worst-case shock is in the 7-9% range, which is recoverable. The 25-30% band is calibrated so the cap fires before the math becomes ruinous, but doesn't fire so early that a portfolio with only 1-2 tagged Sports Cars positions can't operate.

What if a tagged name has structurally lower exposure

The framework's weighting reflects partial dependence. AAPL gets a 50% weight because only silicon is TSMC-dependent (services + iPhone hardware mix dilutes). MU is tagged at lower weight because much of memory production is non-TSMC. INTC isn't tagged at all (foundry-internal). The weighting is calibrated annually based on disclosed manufacturing concentration; the trader doesn't need to memorize it but should know that "tagged" doesn't mean "100% sovereignty risk." The dashboard surfaces the weight on each tagged position.

The three-cap composition rule

The Sovereignty Cap composes with the per-sector cap (L14) and per-sleeve cap (L13). All three apply simultaneously:

NVDA at 15% in Sports Cars sleeve:
  · Per-name cap (L13)        — 15% ≤ 18% sleeve max ✓
  · Sleeve cap (L13)          — Sports Cars 40% ≤ 50% target band ✓
  · Sector cap (L14)          — Tech 32% > 30% ✗  (FAILS)
  · Sovereignty Cap (L21/E10) — tagged 22% ≤ 28% ✓

Result: refused on sector cap, regardless of other gates passing.

Adding a position is approved only when all four checks pass. The most common failure mode is sector cap; the Sovereignty Cap is rarer but catches the cases where the trader has spread tech across non-tech sectors (TSM is "Technology" but in the same exposure bucket as AAPL despite GICS sectoring AAPL differently in some classifications) and accidentally over-concentrated in fab-dependent names.

The real lesson

Most risk pillars are about events that have happened before — drawdowns, earnings gaps, sector rotations, regime shifts. The Sovereignty Cap is the one pillar specifically about an event that hasn't happened yet but would be catastrophic if it did. Insurance for the rare-but-large case. The 25-30% cap isn't predicting Taiwan goes kinetic; it's making sure that if it does, your portfolio survives the repricing. Most retail portfolios overlook this entirely because the base rate is zero in living memory. The framework's design is to bound the exposure regardless of the base-rate question — exactly the kind of insurance the math says you carry on insurable concentration risks.

Related: L13 — sleeve allocation · L14 — sector rotation · L20 — macro regimes