Portfolio Construction

Portfolio Construction (Bubble-Aware, All-Weather)

Builds a target allocation that participates in equity upside but survives a concentration-bubble unwind, without requiring a market-timing call. The thesis: at high valuations + extreme concentration, the fix is de-concentrate the equity core + add uncorrelated diversifiers + keep dry powder, not "predict the crash" or "go to cash."

Mandatory framing

• Educational framework, not personalized advice; consult a fee-only fiduciary at meaningful size.
• Assumes a long-horizon investor treating this as growth capital, with emergency fund and

high-interest debt already handled.

• Defense underperforms in bull markets — that lag is the premium paid to cap the left tail.

The four macro regimes (the design principle)

Never bet everything on one. Hold something that wins in each: prosperity (stocks, small-cap value), recession/deflation (long Treasuries, cash), inflation (gold, commodities, TIPS), tight money (T-bills).

Target allocation — pick a risk tier

Choose by the drawdown the user can actually live through, not by the return they want.

Why this shape (the evidence)

• Equity is de-concentrated (equal-weight/intl/value/min-vol) instead of ~40% AI mega-caps.
• Gold + trend are the diversifiers that worked in 2022 when bonds failed; trend is "crisis alpha."
• Treasuries kept modest (not classic All-Weather's 40%) given the 2022 duration lesson and low-yield start.
• Dry powder earns ~4-5% and is deployed into declines by rule (hand off to dip-tranches-strategy).
• Backtest (this repo's backtests/crash_protection_backtest.py): diversified mixes earned ~7-8%/yr 2000-2026 with

−16% to −24% max drawdowns vs the S&P's −55% / QQQ's −83%, and roughly doubled through the 2000-2009 lost decade.

Sizing mechanics

• Optionally scale the risky sleeves by the exposure_multiplier from regime-detection

(reduce equity/trend, park the difference in T-bills) — keep gold/TIPS as structural diversifiers.

• Within sleeves, default to the listed weights; for the multi-name equity sleeve use the ETF

(don't stock-pick) unless a separate factor skill is in play.

• Respect risk-management caps (per-asset, per-sleeve, gross) — it can scale you down, never up.

Outputs (contract)

{
  "as_of": "2026-05-29",
  "tier": "balanced",
  "targets": {"RSP": 0.18, "VXUS": 0.12, "AVUV": 0.08, "USMV": 0.07,
              "GLD": 0.10, "DBMF": 0.10, "TLT": 0.07, "SCHP": 0.03,
              "SGOV": 0.22, "BTAL": 0.03},
  "equity_beta": 0.45,
  "notes": "Dry powder (SGOV) routed to dip-tranches skill for drawdown deployment."
}

Hand-offs

• Cash deployment schedule + dip reserve → dip-tranches-strategy.
• Drift/rebalance back to these targets → rebalancing.
• Exposure scaling → regime-detection (input) and risk-management (override).
• Loss harvesting on the taxable sleeves → tax-loss-harvesting.