The 13 risk pillars — what each one actually protects against

The 11-point technical score tells you which stocks look good. The 13 risk pillars tell you which stocks you're allowed to own. This second layer is what separates a screener from a trading system — and it's where most retail frameworks stop.

Every ticker Swing Deck audits runs through both layers. Score high on technicals but fail a pillar — you're out. It doesn't matter that NVDA looks like a 95 on the 11-point scan; if the VIX is at 32 and you're already at 28% sector exposure to semis, the pillars will refuse to let you add.

Here's what each pillar checks and — more importantly — what kind of loss each one was designed to prevent. I've lived the failure mode for most of them personally. The framework didn't come from a textbook; it came from watching accounts bleed.

Layer 1 — Position-level pillars (1-7)

These run against every ticker you hold. They're the first line of defense.

Pillar 1 — Armor Cap (15% per asset)

No single position can be more than 15% of the portfolio. Cash positions (USFR, SGOV, etc.) are exempt. The cap is structural — it doesn't matter how high-conviction your thesis is. If your weight calculation would push you over 15%, the sizer refuses.

What it prevents: the "I'm so sure about this one" blow-up. Every experienced trader has at least one year where the right answer was "just hold NVDA" — and then the next year where the right answer was "absolutely not, trim immediately." The cap stops you from drawing the wrong conclusion from a 12-month window.

Pillar 2 — Red-Line (7% portfolio drawdown)

If your portfolio drops 7% from its cost-basis high-water mark, the rebalancer generates a 50%-cash pivot. You don't manually decide — a daemon monitors portfolio value every 60 seconds and if the threshold breaches, it writes the rotation orders.

What it prevents: the death-spiral. 7% is the magic number because the recovery math gets brutal past it: a 7% drawdown needs a 7.5% gain to recover, but a 20% drawdown needs a 25% gain. Catching drawdowns early and going to cash preserves optionality for the next regime.

Pillar 3 — Pre-Market Firewall (S&P futures -1.5%)

If S&P E-mini futures are down more than 1.5% at 7:25 AM MST, entry orders are blocked for the day. You can still exit, you can still raise stops — but new long entries don't clear the gate.

What it prevents: catching falling knives on an open. -1.5% in the overnight isn't a disaster, but it's a specific market signal (institutional selling, overseas contagion, unexpected macro print) and adding exposure into that is a losing ev trade nine times out of ten.

Pillar 4 — Sector Correlation Cap (40% per theme)

No more than 40% of the portfolio in any single sector/theme. Semiconductors, AI infrastructure, defense primes, pharma, etc. The engine tags each ticker by theme at audit time.

What it prevents: the pretend-diversification trap. Holding NVDA + AVGO + AMD + SMCI + MRVL feels diversified because they're different tickers, but they all move as one on a fab-capacity disappointment. 40% caps the sector risk to something recoverable.

Pillar 5 — Hard-Cash Floor (10% USD/T-bills)

At least 10% of the portfolio in USFR, SGOV, or bare cash at all times. The sizer refuses to allocate the last 10% into risk assets.

What it prevents: running out of ammo right when you need it. The best buying opportunities come during max-stress environments — when everyone else is forced to sell. If you're 100% invested, you can't add. If you've got 10% dry powder earning 4-5% on T-bills, you can go hunting when spreads widen.

Pillar 6 — Earnings Proximity Blocker (≤ 3 sessions)

No new swing entries if earnings are within 3 trading sessions. The earnings alert fires separately so you know to tighten or close existing positions.

What it prevents: an 18% gap against you on an EPS miss. Swing trades and earnings bets are fundamentally different risk profiles — one is a technical breakout at 2R risk, the other is a binary outcome at 5-8R risk. Mixing them is how traders discover their sizing was wrong.

Pillar 7 — ATR Stop Discipline (2× ATR)

Every position has a stop-loss at entry minus 2× ATR(14). No exceptions. The engine computes it, the dashboard displays it, the Raise-Stop automation only moves it up and only via ATR-based trailing math.

What it prevents: the mental-stop problem covered in last week's post. If the stop lives in your head, it negotiates. If it lives at the broker and can only ratchet up, it doesn't.

Layer 2 — Defensive calibration (8-13) — activate at VIX > 25

The second set of pillars are dormant during calm markets. They wake up when volatility spikes and add constraints on top of the baseline pillars.

Pillar 8 — Defensive Oil Cap (15% when VIX > 25)

Energy exposure caps at 15% of portfolio in high-VIX environments. Oil + energy names are hard to size correctly in macro-off regimes because they're not behaving as inflation hedge OR as risk asset — they're behaving as both, chaotically.

Pillar 9 — Hardware War Cap (30%)

When the "energy-inversion" regime flags (natgas/oil/uranium inverted term structure), hardware-adjacent exposure (semis + compute infra) caps at 30%. This is a specific macro regime where the leaders of the previous cycle become the laggards of the current one.

Pillar 10 — Velocity Exception (35% sleeve cap)

If a ticker qualifies for the Velocity Track (high-momentum, high-beta, specific technical signature), it can exceed the 25% Sports-Cars sleeve cap up to 35% — but only for that ticker, only while Velocity flag is active, and only if Armor Cash is above 15%.

Pillar 11 — Gap Protection

When gap risk is elevated (earnings cluster weeks, Fed meeting weeks, CPI weeks), position sizes are reduced by 20% across the board. Not a cap — a sizing multiplier.

Pillar 12 — Armor Yield Floor (^IRX ≥ 4.0%)

If the 13-week T-bill yield drops below 4%, the rebalancer rotates cash positions out of USFR/SGOV and into longer-duration fixed income. Sub-4% means the risk-free rate is starting to converge with dividend yields — time to add duration.

Pillar 13 — Diplomatic Decay (ceasefire-adjacent)

The weirdest one. When a geopolitical ceasefire event is within 48 hours (Ukraine, Gaza, Taiwan — any named truce or summit), Sports-Cars (high-momentum tech) scores are penalized by 10 points. This sounds esoteric but has historical precedent: rotation out of defense primes and into risk-on tech happens violently on ceasefire news, and the wrong side of that rotation is expensive.

Why pillars, not rules

Most retail trading frameworks are rules: "buy when the 50-day crosses the 200-day, sell when it uncrosses." Rules are brittle because they're path-dependent — you take the trade because the rule fired, and you feel cheated when it doesn't work.

Pillars are gates. They don't generate trades. They filter out trades your technical scanner wants to take but shouldn't. The distinction matters because it changes your relationship to losses. A rule-based system that loses money feels like betrayal. A pillar-based system that loses money feels like the volatility environment being what it is — the pillars reduced exposure, the losses were smaller than they would have been, and the cash floor means you have ammo left for the next setup.

How Swing Deck enforces them

Every audit cycle (every 5 minutes):

1. The engine computes the 11-point score for each ticker.
2. Each pillar runs against the current position + macro state.
3. Any pillar violation is flagged on the ticker card with the specific pillar number.
4. The sizer refuses allocations that would breach a cap.
5. The rebalancer generates rotation orders for any breach that requires action (Pillar 2, 5, 12).
6. The alerts engine emails you when a pillar transitions — not every tick, but on the edge (e.g., VIX crossing 25, sleeve drift crossing cap+5%, etc.).

You don't have to remember the pillars. You don't have to check them manually. They're the audit's refusal mechanism, not a checklist you run through. When they're working right, you don't notice them — you just don't make the trade you would have made six months ago that would have blown up.