Strategy Explorer VALIDATED
Backtest 2021-11-29→2026-06-01 · 56 monthly decisions · Advisory only
Backtest as of 2026-07-25 — a fixed historical study; it does not auto-update.
The live forward-test updates on the
Portfolio page.
Equity Curve: Strategy vs Basket vs NVDA VALIDATED
— one momentum-regime path; momentum-driven; results must not be extrapolated
Research notes (read before interpreting):- Outperformance is regime-dependent: the strategy UNDERperformed the 2023 V-rebound (kill switch in cash) and outperformed 2024–2026. Result rests on one momentum-bull path / ~56 monthly decisions. Do NOT extrapolate a durable edge.
■ Strategy (equal/realistic)
■ Basket
Pink = kill-switch in cash
Kill-switch fires: 2022-01-25, 2024-07-30, 2025-01-27, 2026-07-07
Per-Year Breakdown — regime dependence made visible
2023 underperformance = kill switch in cash during the V-rebound. This is the honest cost of capital protection.
| Year | Strategy | Basket | Edge |
| 2022 |
-17.7% |
-27.7% |
++10.0% |
| 2023 |
+37.0% |
+70.7% |
-33.7% |
| 2024 |
+86.7% |
+65.3% |
++21.4% |
| 2025 |
+85.0% |
+65.7% |
++19.3% |
| 2026 |
+150.2% |
+70.0% |
++80.2% |
Signal Attribution VALIDATED
— momentum carried the result; val/trend only cut drawdown
| Variant | Total ret | CAGR | Sharpe | Max DD |
| Momentum only |
+912.3% |
+64.5% |
1.50 |
-25.1% |
| Valuation only |
+184.0% |
+25.2% |
0.98 |
-29.1% |
| Trend only |
+355.0% |
+38.5% |
1.25 |
-26.9% |
| Full blend (frictionless) |
+869.5% |
+63.0% |
1.58 |
-22.2% |
| Full blend (realistic) |
+846.8% |
+62.1% |
1.56 |
-22.3% |
All Variants VALIDATED
Friction drag (equal, frictionless→realistic): +22.6% of total return over the full period. A strategy that only wins frictionless is not real.
| Variant | Tot ret | CAGR | Sharpe |
Vol | Max DD | Kill fires |
| Equal/frictionless |
+869.5% |
+63.0% |
1.58 |
+34.9% |
-22.2% |
4 |
| Equal/realistic |
+846.8% |
+62.1% |
1.56 |
+34.9% |
-22.3% |
4 |
| Vol-scaled/frictionless |
+567.8% |
+50.4% |
1.42 |
+32.4% |
-24.5% |
4 |
| Vol-scaled/realistic |
+549.0% |
+49.5% |
1.41 |
+32.4% |
-24.6% |
4 |
| Basket/frictionless |
+499.0% |
+46.9% |
1.28 |
+34.8% |
-37.7% |
0 |
| Basket/realistic |
+496.0% |
+46.8% |
1.28 |
+34.8% |
-37.7% |
0 |
| NVDA buy-hold |
+520.1% |
+48.0% |
1.01 |
+52.3% |
-66.4% |
0 |
Earnings Surprise Study COMMODITY
— avoid-misses; regime-dependent; effective n ≪ raw n
Canonical figure (EXCESS over basket, ≥2021-11-29):
The signal is almost entirely downside — misses lag the basket ~4%/21d; beats merely match it.
Raw returns look positive for everyone because the sector rose ~4%/21d. Welch t≈2.9 but
AI-infra events are correlated, so effective n ≪ raw n. Treat as suggestive, not proven.
| EPS Bucket | n | Mean excess/21d | Interpretation |
| <0% (miss) |
135 |
-4.1% |
miss → lag basket |
| 0–5% |
176 |
-1.2% |
moderate surprise |
| 5–10% |
152 |
-1.0% |
moderate surprise |
| >10% (bigbeat) |
219 |
-0.3% |
bigbeat → ≈match basket |
Per-year sign stability (21d excess)
| Year | Big-beat % positive excess | Miss % negative excess | Stable? |
| 2021 |
0% (n=1) |
— |
no/weak |
| 2022 |
46% (n=35) |
58% (n=45) |
no/weak |
| 2023 |
45% (n=49) |
59% (n=27) |
no/weak |
| 2024 |
36% (n=56) |
84% (n=25) |
no/weak |
| 2025 |
48% (n=56) |
76% (n=29) |
no/weak |
| 2026 |
36% (n=22) |
56% (n=9) |
no/weak |