A systematic risk overlay on the S&P 500. Hold SPY when the trend is intact; step to cash when a sustained selloff confirms. Twelve trades in thirty-three years. No stock picking.
Run SPY's monthly close through a Kalman filter to remove month-to-month noise.
Compute a 9-month moving average of the filtered price. This is the trend line.
Sell SPY for cash if price falls more than two standard deviations below the trend. Buy back when price reclaims the trend.
| Crisis window | VolBand | SPY | VolBand DD | SPY DD |
|---|---|---|---|---|
| 2000-02 Dot-com bust | -22.9% | -36.5% | -23.1% | -41.6% |
| 2008-09 GFC | -9.8% | -47.1% | -10.7% | -50.8% |
| 2011 Euro debt crisis | -16.8% | -7.1% | -16.8% | -16.7% |
| 2015-16 China selloff | -9.8% | -1.0% | -9.8% | -6.7% |
| 2018 Q4 sell-off | -7.6% | -7.6% | -9.3% | -9.3% |
| 2020 COVID crash | -1.4% | -1.4% | -13.0% | -13.0% |
| 2022 bear market | -15.2% | -13.2% | -16.2% | -20.3% |
The strategy's value is concentrated in long drawdowns (2000 dot-com, 2008 GFC). In fast crashes and shallow corrections it tracks SPY or whipsaws slightly. This is a tail-risk overlay, not a market-timing edge.
First trading day of each month. In SPY, or in cash. Most months: no change.
Your brokerage account, your custody. Trade SPY/VOO and SGOV/BIL yourself.
Twelve switches in thirty-three years. Low friction, low tax drag.
Research-only signal service. Not registered investment advice. Past performance does not guarantee future results. Backtest is hypothetical; live results may differ. Not an offer to buy or sell any security.