iM-DrawdownShield holds the S&P 500 ETF (SPY) or, when its published rule signals an exit, an escape asset. Two escape choices are shown here:
- Cash: 1–3 month T-bills (BIL).
- Best ETF: on the exit day, the highest bull-rated of the Treasury ETFs SHY, IEF and TLT, held until the model re-enters SPY; T-bills if none of the three is rated bullish.
All figures run from February 9, 1995 to October 30, 2026. Trades are executed at the close of the day on which the switching price published the previous evening is crossed. A signal that arrives without a prior alert is executed at the next day’s close.
Returns and deepest falls
| Return per year | Deepest fall | |||||
|---|---|---|---|---|---|---|
| Period | SPY | Cash | Best ETF | SPY | Cash | Best ETF |
| 3 years | 23.3 | 20.9 | 21.2 | −18.8 | −10.0 | −10.1 |
| 5 years | 12.3 | 13.3 | 13.5 | −24.5 | −12.9 | −12.9 |
| 10 years | 15.3 | 16.2 | 16.5 | −33.7 | −13.1 | −12.9 |
| 1995–2026 | 11.0 | 12.5 | 13.2 | −55.2 | −25.1 | −19.0 |
| In-sample 2002–2019 | 8.0 | 10.0 | 10.6 | −55.2 | −17.1 | −17.6 |
| Out-of-sample 1995–2001 | 15.2 | 14.9 | 16.0 | −35.5 | −19.0 | −19.0 |
| Out-of-sample 2020–2026 | 14.9 | 16.9 | 17.2 | −33.7 | −13.1 | −12.9 |
Per cent. 60 switches in total, 1.9 per year; time in SPY 84%; median spell in SPY 124 trading days, out of SPY 33. The rule’s constants were set on 2002–2019 only; 1995–2001 and 2020–2026 were held back.
Risk-adjusted measures
| Sharpe ratio | Beta to SPY | Alpha per year | |||||
|---|---|---|---|---|---|---|---|
| Period | SPY | Cash | Best ETF | Cash | Best ETF | Cash | Best ETF |
| 1 year | 0.60 | 0.60 | 0.60 | 1.00 | 1.00 | +0.0 | +0.0 |
| 3 years | 1.17 | 1.25 | 1.27 | 0.65 | 0.65 | +3.9 | +4.2 |
| 5 years | 0.55 | 0.73 | 0.74 | 0.59 | 0.59 | +4.1 | +4.3 |
| 10 years | 0.75 | 1.04 | 1.06 | 0.53 | 0.52 | +6.4 | +6.8 |
| 1995–2026 | 0.52 | 0.77 | 0.81 | 0.50 | 0.49 | +5.4 | +6.1 |
| In-sample 2002–2019 | 0.44 | 0.75 | 0.79 | 0.42 | 0.40 | +5.6 | +6.3 |
| Out-of-sample 1995–2001 | 0.58 | 0.64 | 0.70 | 0.70 | 0.70 | +2.4 | +3.5 |
| Out-of-sample 2020–2026 | 0.66 | 0.98 | 1.00 | 0.50 | 0.50 | +7.3 | +7.6 |
All measures on daily returns in excess of T-bills (BIL). Alpha in per cent per year against SPY, whose beta is 1 and alpha 0 by definition. Over the last year the model was in SPY throughout, so its figures equal SPY’s. Sortino ratio 1995–2026: SPY 0.74, Cash 1.08, Best ETF 1.14.
Major declines, SPY peak to trough
| Decline | Dates | SPY | Cash | Best ETF |
|---|---|---|---|---|
| Asian/LTCM crisis | 7/17/98 – 10/8/98 | −18.2 | −15.2 | −14.0 |
| Dot-com bear | 3/24/00 – 10/9/02 | −47.5 | −13.3 | −5.4 |
| Global financial crisis | 10/9/07 – 3/9/09 | −55.2 | −14.1 | −9.2 |
| 2011 debt ceiling | 4/29/11 – 10/3/11 | −18.6 | −11.6 | −11.5 |
| 2015–16 slowdown | 5/21/15 – 2/11/16 | −13.0 | −11.4 | −7.2 |
| Q4 2018 | 9/20/18 – 12/24/18 | −19.3 | −5.1 | −4.4 |
| COVID crash | 2/19/20 – 3/23/20 | −33.7 | −12.2 | −11.4 |
| 2022 inflation bear | 1/3/22 – 10/12/22 | −24.5 | −7.3 | −7.3 |
| 2025 tariff shock | 2/19/25 – 4/8/25 | −18.8 | −9.8 | −9.6 |
Total return over each SPY decline, per cent.
Calendar years
| Year | SPY | Cash | Best ETF | Best ETF vs SPY |
|---|---|---|---|---|
| 1996 | 22.6 | 22.6 | 22.6 | 0.0 |
| 1997 | 33.5 | 33.5 | 33.5 | 0.0 |
| 1998 | 28.7 | 20.6 | 22.7 | −6.0 |
| 1999 | 20.4 | 20.4 | 20.4 | 0.0 |
| 2000 | −9.7 | −14.0 | −12.6 | −2.9 |
| 2001 | −11.8 | −3.4 | 0.0 | +11.8 |
| 2002 | −21.6 | −2.7 | 0.3 | +21.9 |
| 2003 | 28.2 | 23.2 | 23.0 | −5.2 |
| 2004 | 10.7 | 10.7 | 10.7 | 0.0 |
| 2005 | 4.8 | 0.1 | −0.1 | −4.9 |
| 2006 | 15.8 | 11.2 | 11.1 | −4.7 |
| 2007 | 5.1 | 0.2 | 0.8 | −4.3 |
| 2008 | −36.8 | −4.3 | 1.4 | +38.2 |
| 2009 | 26.4 | 39.1 | 38.3 | +11.9 |
| 2010 | 15.1 | 10.0 | 10.9 | −4.2 |
| Year | SPY | Cash | Best ETF | Best ETF vs SPY |
|---|---|---|---|---|
| 2011 | 1.9 | −6.2 | −8.4 | −10.3 |
| 2012 | 16.0 | 16.0 | 16.0 | 0.0 |
| 2013 | 32.3 | 32.3 | 32.3 | 0.0 |
| 2014 | 13.5 | 13.5 | 13.5 | 0.0 |
| 2015 | 1.3 | −5.2 | −5.1 | −6.4 |
| 2016 | 12.0 | 8.4 | 11.4 | −0.6 |
| 2017 | 21.7 | 21.7 | 21.7 | 0.0 |
| 2018 | −4.6 | 5.3 | 6.2 | +10.8 |
| 2019 | 31.2 | 19.6 | 19.4 | −11.8 |
| 2020 | 18.4 | 26.7 | 28.1 | +9.7 |
| 2021 | 28.7 | 28.7 | 28.7 | 0.0 |
| 2022 | −18.2 | −6.7 | −6.7 | +11.5 |
| 2023 | 26.2 | 16.2 | 16.4 | −9.8 |
| 2024 | 24.9 | 24.9 | 24.9 | 0.0 |
| 2025 | 17.7 | 17.3 | 18.0 | +0.3 |
Total return, per cent. “Best ETF vs SPY”: difference in percentage points; green = Best ETF ahead of SPY, red = behind, grey = fully invested all year (equal to SPY). 1995 (from February 9) and 2026 (to October 30): always invested, SPY and both variants +30.2 and +11.7.
| Full years 1996–2025 | SPY | Cash | Best ETF |
|---|---|---|---|
| Losing years | 6 | 7 | 5 |
| Years worse than −10% | 4 | 1 | 1 |
| Worst year | −36.8 | −14.0 | −12.6 |
| Years better than SPY | — | 7 | 8 |
| Years equal to SPY (always invested) | — | 10 | 10 |
Returns rounded to 0.1%.
Disclosure. All performance figures are hypothetical results of applying the published rules to historical data. Escape-asset prices before each ETF’s launch are simulated. The figures do not represent actual trading and do not include costs, slippage or taxes. Past performance, actual or simulated, does not guarantee future results. iM-DrawdownShield is a mathematical model; it is not investment advice.