Gold/SPX Switching Replication

Study: Gold/SPX Switching Replication

Code: https://github.com/receiptsandregimes/receipts-lab/tree/main/studies/gold-spx-switching-replication (status: protocol-only, seeded 2026-10-12)

Question

Can a plausible reconstruction of returnstream’s gold-spx-switching-strategy reproduce CAGR ≈ 12.7% / max DD ≈ 31% (1977–2026) and positive returns in the dot-com and subprime windows — and how sensitive are those numbers to the undisclosed parameters?

Claims tested

• gold-spx-switching-outperforms-both-assets
• switching-positive-in-dotcom-subprime
• gold-trend-filter-underperforms-bnh

Data & assumptions

• Universe: gold spot per lbma-gold-price composite leg (frozen LBMA PM-fix history 1968-04→2026-02 + seam-normalized GLD total-return live extension, GC=F month-ends second opinion — the FRED GOLDPMGBD228NL named at proposal time is dead; see log 2026-10-12), S&P 500 total return (or price index + reinvested dividend proxy), 3-month T-bill (FRED TB3MS), 1977-01-01 → present, monthly bars.
• Costs: base run costless (matches vendor style), sensitivity run at 10 bps per switch; no leverage, fully invested.
• Engine: self-contained Python (pandas) under studies/gold-spx-switching-replication/code/.
• No signal-peeking: positions formed on prior month-end data, executed next month.

Method

Because the exact rules are paywalled, grid a family of reconstructions:

1. Relative-strength lookback L ∈ {3, 6, 9, 12} months; hold asset with higher trailing return.
2. Risk-free filter variants: (a) both trailing returns < T-bill trailing return → cash; (b) both < 0.
3. Baselines: gold B&H, SPX B&H, gold-only 1-yr trend filter (tests claim 3). Report CAGR, max DD, Sharpe per variant; stress-window returns for Mar-2000–Oct-2002 and Nov-2007–Mar-2009; heatmap of CAGR/DD across L × filter to expose parameter sensitivity. Compare best/median variant against vendor’s 12.7% / 31%.

Questions this study must answer or update

• gold-trend-filter-post-2023 (baseline leg answers the gold half)
• does-the-cash-leg-earn-its-keep (cash-filter variants vs no-cash baseline)
• both-assets-down-regimes (cash variants are the strategy’s answer to this — measure it)

Results

(pending — no code run yet)

Verdict

(pending)

Next tests prompted

• If robust: compare against SPX-only trend + cash (does gold leg add anything?).
• Weekly execution variant to check monthly-bar smoothing effects.
• If Frank buys the ebook, rerun with the true rule set and diff.
• Cross-link: gold-spx-ratio-rotation-replication tests the fully-disclosed no-cash-leg sibling (gold-spx-rs-ratio-rotation) — run both and compare whether the T-bill escape hatch earns its keep.