The 10-month moving average strategy holds an index only when its month-end close sits above its 10-month simple moving average (SMA), and moves to cash when it does not. Run on $SPY from late 1993 through October 2026, it compounded at 9.42% a year versus 10.74% for buy-and-hold, but it cut the worst month-end drawdown from -50.78% to -22.96%.
That trade, less return for far less pain, is the whole story. Only $QQQ beat buy-and-hold on return, and it did so because the rule sat out most of the 2000–2002 Nasdaq collapse.
At the October 30, 2026 month-end close, all four ETFs tested ($SPY, $QQQ, $IWM and $DIA) carried an "invested" signal for November. The interim read at the November 18 close is far shakier. Details below.
Where the 10-month rule comes from
The rule was popularized by Mebane Faber's 2007 paper, "A Quantitative Approach to Tactical Asset Allocation," published in the Journal of Wealth Management and heavily downloaded from SSRN. As WealthManagement.com described the method: at the end of every month, the investor checks each asset's price against its 10-month moving average, keeps what sits above it and moves what sits below it to cash.
The appeal is plain. One check a month. No intraday screens, no gut calls. Ten month-ends span roughly 210 trading sessions, so the line is a slower, monthly cousin of the 200-day average that drives the golden cross and death cross signals.
The exact rules tested
- Take the close on the last trading day of each month. This test uses the Yahoo Finance adjusted close, which folds in dividends and splits, so returns approximate total return.
- Average the latest 10 month-end values. That is the 10-month SMA.
- If the month-end close is above the SMA, hold the ETF for the next month. If not, hold cash earning 0%.
- Trade at the month-end close, with no commissions, taxes or slippage.
Each test starts at the first month-end with 10 months of history: October 1993 for $SPY (its data begins January 29, 1993), October 1998 for $DIA, December 1999 for $QQQ and February 2001 for $IWM. Every test ends at the October 30, 2026 month-end, and buy-and-hold is measured over the identical window.
Worked example: the October 2026 $SPY signal
Here are $SPY's ten month-end adjusted closes from January through October 2026: 686.61, 680.67, 647.06, 715.04, 752.67, 744.92, 745.18, 765.15, 762.63 and 755.72.
Their average is about 725.56. The October 30 close of 755.72 sits 4.16% above that line (755.72 ÷ 725.56 − 1). Signal: invested for November.
Now look at March. The March 31 close of 647.06 sat 1.52% below its 10-month SMA of 657.05, so the rule went to cash for April 2026. $SPY then gained 10.51% in April. $QQQ was flagged out the same month and missed a 15.69% April gain. That is the rule's cost in one line.
Backtest results: $SPY, $QQQ, $IWM and $DIA
All figures use Yahoo Finance adjusted month-end closes. Max drawdown is measured on month-end values, so intra-month lows were deeper. "Worst miss" is the biggest monthly gain buy-and-hold earned while the rule sat in cash.
| ETF | Return months | CAGR: rule | CAGR: buy-hold | Max DD: rule | Max DD: buy-hold | Switches | Time invested | Worst miss |
|---|---|---|---|---|---|---|---|---|
| $SPY | Nov 1993–Oct 2026 | 9.42% | 10.74% | -22.96% | -50.78% | 46 | 77.8% | +12.70% (Apr 2020) |
| $QQQ | Jan 2000–Oct 2026 | 8.92% | 8.52% | -36.54% | -81.08% | 50 | 72.0% | +18.48% (Oct 2002) |
| $IWM | Mar 2001–Oct 2026 | 5.71% | 8.58% | -31.25% | -52.47% | 51 | 70.1% | +15.39% (Apr 2009) |
| $DIA | Nov 1998–Oct 2026 | 6.32% | 8.88% | -31.34% | -47.03% | 52 | 76.8% | +14.03% (Oct 2022) |
In dollars, $10,000 in the $SPY rule grew to $195,148, against $289,437 for buy-and-hold. The rule ran with less volatility, though: 10.74% a year versus 14.85% (monthly returns, annualized).
$QQQ is the exception. The rule went to cash on the September 29, 2000 close and stayed out through February 2003, a stretch in which buy-and-hold fell 71.65%. That exit did the heavy lifting: $10,000 grew to $98,898 under the rule versus $89,772 for buy-and-hold.
The drawdown math explains why traders put up with the lag. A -50.78% hole needs a 103.2% gain to recover; a -22.96% hole needs 29.8%. The full arithmetic is in the TOT drawdown recovery math breakdown.
Why it lags: whipsaws
Most exits cost money. $SPY had 23 separate cash spells, and buy-and-hold rose during 19 of them. The rule earns its keep from a handful of big ones: the spell from the December 31, 2007 signal through June 2009 sidestepped a 34.77% buy-and-hold decline, including October 2008, the worst $SPY month in the sample at -16.52%.
2022 shows the downside. $SPY was invested for January, February, April and December, and every one of those months fell. It sat in cash for March, which rose 3.76%. The rule finished 2022 down 20.97% versus 18.18% for buy-and-hold, and that choppy stretch produced its own worst drawdown of -22.96%, from December 2021 to February 2023.
The monthly check is also a blind spot. The rule cannot react inside a month, so a crash that starts on the second trading day of a month can be fully absorbed before the next check.
The signal now: October 30 and an interim November 18 read
The official signal for November was set at the October 30, 2026 close. The interim column swaps in the November 18 close for the unfinished November month-end, which drops January 2026 from the 10-month window. It is a snapshot, not a signal: November is not over.
| ETF | Oct 30 close | 10-month SMA | Gap | Signal for Nov | Nov 18 close | Interim gap |
|---|---|---|---|---|---|---|
| $SPY | 755.72 | 725.56 | +4.16% | Invested | 730.20 | +0.04% |
| $QQQ | 690.49 | 677.32 | +1.95% | Invested | 659.51 | -3.20% |
| $IWM | 279.33 | 277.04 | +0.83% | Invested | 270.12 | -2.93% |
| $DIA | 517.13 | 502.83 | +2.84% | Invested | 501.73 | -0.57% |
At the November 18 close, $QQQ, $IWM and $DIA sat below their interim 10-month lines, and $SPY sat just 0.04% above. Under the rule, nothing changes until the November month-end close. The table makes no forecast; it only shows where prices stood.
These tests ignore commissions, taxes and slippage, fill every signal at the month-end close, pay 0% on cash, cover only ETFs that survived, and rely on one data source (Yahoo Finance), whose month-end closes matched Nasdaq.com on every month-end of the overlapping 10 years. More rule-based studies are in the Analysis section at Traders On Trend.
Bull/Bear Verdict
Bull Case: Where it helps: the 10-month rule cut the worst month-end drawdown by a third to more than half on all four ETFs, from -50.78% to -22.96% on $SPY and from -81.08% to -36.54% on $QQQ, with one decision a month and about 1.4 switches a year on $SPY.
Bear Case: Where it misleads: it lagged buy-and-hold on return for $SPY, $IWM and $DIA, most of its exits were whipsaws (19 of 23 $SPY cash spells saw prices rise), and it missed rebounds such as April 2026's 10.51% $SPY gain.
FAQ
Does the 10-month moving average strategy beat buy-and-hold?
On return, rarely. In this test only $QQQ came out ahead (8.92% versus 8.52% a year). The consistent edge was risk: lower volatility and a far smaller worst drawdown on all four ETFs.
Should the moving average use price or total return?
This test uses the dividend-adjusted close. On a price-only basis, the October 30 signals were unchanged: $SPY sat 3.77% above its price-only 10-month SMA of 728.26, and all four ETFs stayed invested.
Is the 10-month SMA the same as the 200-day moving average?
Close, not identical. Ten month-ends cover roughly 210 sessions, but the 10-month rule averages only 10 data points and is checked once a month, while the 200-day line averages every session and can flip on any day.
How often does the 10-month rule trade?
$SPY switched position 46 times in 33 years, about 1.4 times a year: 23 exits and 23 re-entries. $IWM traded most often, with 51 switches in about 25.7 years, roughly two a year.