Simulation only — engine not reporting
The strategy: the dip maker (base tier) plus a PERMANENT GLDx vault — no rungs, the gold converts once at initialization and never deploys. This is the study's "dip + gold, no rungs" benchmark (purple line below) run live as the A/B against the reserve strategy. Backtests below use SPY/GLD as proxies at live-measured costs (24bp net/cycle, 10bp/side GLDx). Full study: reports/2026-08-05/gld-reserve-study.md.
Growth of $1 — 2016-2026 (minute)
the well-measured decade — real cycling rate, includes COVID + 2022. Dashed grid line at 1.0x = holding USDC.
+11.70%/yr
dd -22.9%
+8.61%/yr
dd -34.4%
+13.17%/yr
dd -33.3%
+12.94%/yr
dd -27.8%
+10.97%/yr
dd -21.3%
Growth of $1 — 2000-2026 (daily)
adds dot-com + GFC; daily bars undercount cycles ~4.5x, crash shapes are the point. Dashed grid line at 1.0x = holding USDC.
+7.46%/yr
dd -30.9%
+1.45%/yr
dd -55.3%
+6.33%/yr
dd -56.5%
+10.56%/yr
dd -45.3%
+8.07%/yr
dd -35.7%
Crash windows — 2016-2026 minute
| window | Dip (current) | Dip + GLD | SPY hold | deep cycles |
|---|---|---|---|---|
| COVID 2020 | +9.7% | +21.8% | +15.0% | 25 |
| 2022 grind | -6.7% | -0.7% | -8.1% | 6 |
| gold rush 24-26 | +38.2% | +59.9% | +56.1% | 6 |
Crash windows — 2000-2026 daily
| window | Dip (current) | Dip + GLD | SPY hold | deep cycles |
|---|---|---|---|---|
| dotcom 00-03 | -24.7% | +8.3% | -27.0% | 8 |
| GFC 07-09 | -25.9% | +14.2% | -27.1% | 9 |
| COVID 2020 | +3.1% | +15.4% | +14.8% | 6 |
| 2022 grind | -5.0% | +7.1% | -8.3% | 8 |
| gold rush 24-26 | +7.7% | +62.8% | +54.3% | 3 |
How much GLD per $1 of machine
| reserve | 2016-26 CAGR | maxDD | 2000-26 CAGR |
|---|---|---|---|
| 0 (current) | +8.61% | -34.4% | +1.45% |
| 0.5x | +10.78% | -27.0% | +6.04% |
| 1x (recommended) | +11.73% | -23.9% | +7.25% |
| 2x | +12.60% | -21.3% | +8.19% |
| 3x | +13.02% | -21.1% | +8.59% |
The rungs eventually deploy the reserve's entire then-current value in a real crash — beyond 1x the extra return is mostly just owning more gold, so 1x captures the full mechanism and caps the gold bet at half the book.
Where the new strategy's edge comes from (2016-26, 1x)
Every gold series obtainable starts at the 2000 generational bottom — all measured gold carry is bull-regime carry. 1980-2000 gold lost ~65% and cannot be backtested against a SPY crash. The rungs are strategy; the gold is a position.
Parked until scale — studied, approved, waiting for the numbers