Trend filters on 18 lazy portfolios: drawdowns, returns and whipsaws
| Portfolio | Comparison window starts | Max drawdown, plain → 10-month filter | CAGR, plain → 10-month filter | Months in cash | Switches per year | Read |
|---|---|---|---|---|---|---|
| Golden Butterfly | Jan 2006 | -17.8% → -8.4% | 7.7% → 7.2% | 15% | 0.7 | Shallower drawdowns, lower return |
| All Weather | Jan 2006 | -21.5% → -12.7% | 6.6% → 5.6% | 18% | 1.2 | Shallower drawdowns, lower return |
| Permanent Portfolio | Jan 2006 | -15.9% → -8.8% | 7.1% → 6.0% | 20% | 1.2 | Shallower drawdowns, lower return |
| Ivy Portfolio | Jul 1997 | -43.1% → -16.3% | 7.5% → 6.5% | 22% | 1.6 | Shallower drawdowns, lower return |
| Faber GAA | Jan 2006 | -24.9% → -9.3% | 6.4% → 5.8% | 18% | 1.1 | Shallower drawdowns, lower return |
| Three-Fund Portfolio | Jun 1997 | -37.2% → -14.4% | 7.1% → 6.7% | 23% | 1.0 | Shallower drawdowns, lower return |
| Wealthfront Classic | Aug 2001 | -47.0% → -16.7% | 8.5% → 7.7% | 24% | 1.4 | Shallower drawdowns, lower return |
| Betterment Core | Aug 2001 | -50.2% → -14.4% | 8.5% → 7.9% | 25% | 1.2 | Shallower drawdowns, lower return |
| 60/40 Portfolio | Feb 1988 | -30.3% → -14.3% | 9.1% → 7.6% | 18% | 1.3 | Shallower drawdowns, lower return |
| 40/60 Portfolio | Feb 1988 | -18.8% → -8.7% | 7.9% → 6.5% | 16% | 1.5 | Shallower drawdowns, lower return |
| Yale Endowment | Aug 2001 | -39.5% → -13.2% | 8.0% → 6.9% | 21% | 1.4 | Shallower drawdowns, lower return |
| Core Four | Jul 1997 | -44.4% → -15.8% | 7.9% → 7.7% | 23% | 1.0 | Shallower drawdowns, same return |
| Couch Potato | Aug 2001 | -27.0% → -10.7% | 7.2% → 6.2% | 20% | 1.2 | Shallower drawdowns, lower return |
| Coffeehouse | Jul 1997 | -34.2% → -15.0% | 7.1% → 5.9% | 20% | 1.2 | Shallower drawdowns, lower return |
| No-Brainer | Jun 1997 | -40.9% → -14.9% | 7.6% → 6.8% | 24% | 1.3 | Shallower drawdowns, lower return |
| Larry Portfolio | Jun 1997 | -19.0% → -12.0% | 5.9% → 5.1% | 15% | 1.2 | Shallower drawdowns, lower return |
| Buffett 90/10 | Dec 1992 | -46.3% → -20.7% | 10.2% → 8.9% | 21% | 1.4 | Shallower drawdowns, lower return |
| S&P 500single-fund reference | Mar 1986 | -50.8% → -26.4% | 11.0% → 8.5% | 22% | 1.5 | Shallower drawdowns, lower return |
Each row is the 10-month filter over the months for which the 6-, 10- and 12-month filters all have data. Cash earns 0%, as in the builder. Open a row for the 6- and 12-month columns, the S&P 500's major declines, 2022, quick reversals and the same filter with cash in short Treasuries. The S&P 500 row is the single-fund reference; the same rule on one fund is on the trend-timing strategy page.
Not included: Dragon (starts 2017-05: under 15 years once the 12-month filter's warm-up is removed (and its DBMF and CAOS proxies are themselves category stand-ins)). Every page needs at least 15 years of common history.
How the filter works
The filter looks at the whole portfolio's own value at each month-end. If it is above its average of the last N month-ends the portfolio is held for the next month, otherwise it sits in cash (earning 0%). N is 10 for the headline column, 6 and 12 for the sensitivity columns. It is the whole-portfolio overlay the builder's Add Overlay applies, computed with the same engine function.
Why these three windows
10 months is the window Mebane Faber published in 2007 and the monthly equivalent of the 200-day moving average. 6 and 12 months bracket it. We fixed these three before computing any page and use the same three for every portfolio; we did not search for the window that looks best on each one. Choosing the best of many windows after the fact makes any filter look better than it would have been in real time.
Related
Fund histories: how far back each ETF can be tested · Portfolio strategy guides · All portfolios
Methodology
- Source
- Yahoo Finance via MarketHeist, monthly adjusted close (dividends reinvested), to the last complete calendar month; proxy-extended history as on each fund's history page.
- Numbers
- Every figure is computed by the MarketHeist portfolio engine; no figure is written by hand and no language model is involved. The sentences on each page are rule-based over those numbers. The filter windows are 10 (headline), 6 and 12 months, fixed for the whole set.
Historical simulation, not investment advice.