Simple moving-average regime filters owe almost all of their reputation to two historical episodes: 2001 to 2002 and late 2008. In slow, multi-quarter cascades with low mean reversion, exiting to Treasury bills below the 200-day simple moving average (SMA_200) avoids deep drawdowns.
Modern institutional market structure rarely moves in smooth, uncontested straight lines. Between January 3, 2022 (the exact bull-market peak prior to the Federal Reserve tightening cycle) and November 5, 2026, we tested a standard 2-Gear SMA_200 cash-timing overlay head-to-head against our unlevered 1.00x cash-funded portable-alpha core across 1,216 out-of-sample trading days.
| Portfolio Configuration (2022-2026 OOS) | Cumulative Return | Regime Switches | Primary Drag / Driver |
|---|---|---|---|
| Starling 1.00x Portable-Alpha Core | +61.76% (10.48% CAGR) | 0 binary cash dumps | Residual filing alpha (+7.43% spread over SPY) with stock-level statutory vetoes. |
| S&P 500 Benchmark (SPY Total Return) | +54.33% | Buy and hold | Passive cap-weighted beta (-25.36% max drawdown). |
| 2-Gear SMA_200 Cash-Timing Overlay | +38.45% | 38 switches | -23.31% whipsaw drag from V-bottom re-entry lag and range-bound chop. |
Across 38 separate crossings around the 200-day trend line in 2022, late 2023, and 2025 tariff/rate volatility, binary cash switching repeatedly sold into local panic lows and bought back days later after violent V-shaped short-covering rallies had already repriced the index higher. That empirical result aligns directly with Kaminski & Lo (2014) and Lo & Remorov (2017): when return serial correlation is negative or V-shaped at short horizons, mechanical trend-stop overlays lock in permanent whipsaw losses.
Where Permanent Capital Loss Actually Happens (The 555 Dead Tickers)
When you test strategies on a survivorship-free universe of 3,607 historical Russell 3000 CIKs that retains 555 delisted and bankrupt companies (Silicon Valley Bank, Bed Bath & Beyond, First Republic, Signature Bank, WeWork) with a -30% Shumway delisting haircut, a clear pattern emerges. Broad S&P 500 drawdowns recover; terminal balance-sheet implosions in individual equities do not.
Naive value, accrual, and dip-buying screens look spectacular when backtested exclusively on companies that survived to 2026. Once the 555 delisted companies are restored to the historical panel, unguarded dip-buying strategies suffer catastrophic left-tail losses by stepping in front of insolvency spirals.
Stock-Level Statutory Vetoes Beat Macro Guesswork
Instead of trying to guess S&P 500 macro tops with a lagging moving average, Starling Quantitative stays structurally invested while enforcing hard, deterministic one-way kill-switches at the individual issuer level:
Form 8-K Item 4.02 (Non-Reliance on Financial Statements): When a board or external auditor formally discloses that prior 10-K or 10-Q financial statements can no longer be relied upon, bullish technical or valuation signals are immediately overridden to NO_TRADE.
Form NT 10-K and NT 10-Q (Rule 12b-25 Late Filings): Distressed issuers frequently drop late-filing notices after 16:00 ET on Fridays when audit sign-off breaks down or debt covenants breach. Treating Form NT as an automatic long veto cuts off falling knives weeks before formal Chapter 11 petitions.
Acute Credit-Spread Impulses (HYG/IEF) and Hysteresis Guards: When 5-day high-yield credit spreads widen sharply, long entries on levered balance sheets are gated automatically. And for live position management, requiring a 10-trading-day minimum regime dwell window alongside 2-sweep intraday stop confirmation prevents single-sweep opening-cross illiquidity wicks from shaking out fundamentally sound positions.
Read the Full 3,607-CIK Attribution Study
Explore Research Note RN-01 (DOI: 10.5281/zenodo.24222468) for our 555-ticker delisted cohort methodology, and Research Note RN-03 (DOI: 10.5281/zenodo.24222816) for the complete 1,216-day walk-forward attribution tables: