Quantum BiasMoving Average Strategy Tested: The 10-Month Rule on the Nasdaq
The simplest trend rule in finance, put to work on the index NQ traders follow.
In this article
It cut the worst loss from −75% to −25% and beat buy-and-hold over the full period, mostly by sitting out the 2000–2002 collapse. But since 2013, it's trailed buy-and-hold by more than four points a year. It's a risk tool first, and a return tool only some of the time.
| Question | Data | Answer |
|---|---|---|
| Does Faber's 10-month rule improve on holding the Nasdaq-100? | QQQ monthly total returns, Oct 2000 – Sep 2026, 312 months | Over the full period, yes: 10.9% vs 9.2% a year, worst loss −25% vs −75%. Since 2013, no: 16.0% vs 20.3% |
The rule we tested
In 2007, Mebane Faber published A Quantitative Approach to Tactical Asset Allocation in the Journal of Wealth Management. The working-paper version went on to become one of the most downloaded papers on SSRN. Its timing rule is simple on purpose:
- At each month-end, compare the price with its average over the last 10 month-ends.
- If the price is above the average, hold the asset next month.
- If it's below, hold cash (Treasury bills) next month.
Faber applied it to a diversified basket of asset classes. In the February 2013 update of the paper, covering 1973 to 2012, he reports that timing brought volatility down to single digits and cut the worst drawdown from 46% to less than 10%.
A 10-month average is close to the 200-day moving average a lot of traders watch on daily charts. We applied the rule to just one asset, the Nasdaq-100, through QQQ.
What we found
| Annual return | Volatility | Worst drawdown | |
|---|---|---|---|
| Buy and hold QQQ | 9.2% | 22.2% | −74.6% |
| 10-month rule, before costs | 10.9% | 13.0% | −25.4% |
| 10-month rule, 0.10% cost per switch | 10.7% | — | — |
Exact values from table 2 in this article. Bar lengths share a zero baseline; no new calculation or simulated price path. Read the table and its stated sample and limits in this article.
Over 26 years, the rule was invested 71% of the time and switched 47 times, under two switches a year. Costs barely mattered.
The headline numbers look good. The split by period tells a more useful story.
Two very different halves
| Period | Buy and hold | 10-month rule | Worst drawdown, buy and hold | Worst drawdown, rule |
|---|---|---|---|---|
| 2000–2012 | −2.0% a year | +5.5% a year | −74.6% | −25.4% |
| 2013–2026 | +20.3% a year | +16.0% a year | −32.6% | −18.7% |
From 2000 to 2012, the rule did exactly what trend rules are built for. It stepped aside during the 2000–2002 collapse and most of 2008, turning a losing decade for buy-and-hold into a small gain.
From 2013 to 2026, in a long bull market, the rule paid a price. Every correction that dipped below the 10-month average sent it to cash, and it often got back in higher. It still cut the worst drawdown nearly in half, but it gave up more than four points of return a year to do it.
2022 shows both sides. QQQ lost 32.6% that year. The rule lost 6.8%, because it moved to cash early in the drop.
Where the rule hurt: whipsaws
A trend rule has one weakness that keeps coming back. When the market chops sideways around its average, the rule sells after a drop and buys back after a rise, losing a little each time. Traders call that a whipsaw.
On QQQ, whipsaw years were easy to spot by counting switches. Most years had zero to two. A few had a lot more:
| Year | Switches | What the market did |
|---|---|---|
| 2016 | 6 | Sharp drop in January–February, then a choppy recovery |
| 2004, 2005, 2010 | 4 each | Sideways or corrective stretches around the average |
In those years, the rule's cost wasn't one big loss but a string of small ones: out after a dip, back in a few percent higher. That's the price for being out of 2000–2002 and most of 2008. No version of the rule avoids both. A slower average whipsaws less but gets out of crashes later. A faster one gets out sooner but whipsaws more.
The math behind the result
How did a rule that was only invested 71% of the time earn more than holding all the time? Because of what it avoided. Losses compound unevenly: a −75% drop needs a +294% gain to break even. Cutting the worst drop to −25% means you only need +34% to recover. Over the full sample, the rule's lower return in good years was outweighed by not having to climb out of the deepest hole.
Since 2013, there was no deep hole to avoid. The worst buy-and-hold drop was −32.6%, and the market came back from it fast. In that kind of market, the rule's insurance cost more than it paid.
Why this result is honest, and what it isn't
This is a replication of a published rule, with the rule locked in before we looked at the data. We didn't test 9 or 11 months, or other averages, and pick the best. That matters. In What Do We Know About the Profitability of Technical Analysis? (Journal of Economic Surveys, 2007), Cheol-Ho Park and Scott Irwin reviewed 95 modern studies and flagged data snooping, trying a bunch of rules and keeping the winners, as one of the main reasons published results overstate what traders actually get.
It's also a monthly rule. It says nothing about trading NQ intraday.
What this means if you trade NQ
As a regime filter, not a signal. Whether the Nasdaq-100 is above or below its 10-month average is a simple, public read on the long-term trend. In our test of big down weeks, a similar trend filter, the 40-week average, separated drops that bounced from drops that didn't.
As a sizing input. You could keep full size when the index is above its long-term average and cut it below, without changing any intraday setup.
Not as a promise. The rule's best years were crashes. In a steady bull market, it costs you return.
How to check the rule yourself
- On the last trading day of each month, note QQQ's close.
- Average the last 10 of those closes.
- If the close is above the average, the rule is "in". Otherwise it's "out".
- On a monthly chart, a 10-period simple moving average shows the same thing.
At the end of September 2026, QQQ closed at 739.77, about 10% above its 10-month average of 670.76, so the rule was "in". That's a description of the trend, not a forecast: the state only gets checked again at the next month-end, which is part of why people like the rule.
Limits
- One asset. Faber's paper used several asset classes. Trend rules usually work better across many markets than on one.
- Monthly data and closing prices. Real execution would happen at the next open or the next month's first close.
- No taxes. Switching to cash creates taxable gains in most accounts.
- Starting point. Our test starts in October 2000, early in the 2000–2002 drop, which favors any rule that sat out the crash.
How we did it
- QQQ monthly prices from Alpha Vantage (
TIME_SERIES_MONTHLY), adjusted by us for the two-for-one split of March 20, 2000; dividends from Alpha VantageDIVIDENDS. - Cash earns the 3-month Treasury yield, monthly average (Alpha Vantage
TREASURY_YIELD). - Test period: October 2000, the first month with a full 10-month average in our data, to September 2026.
- Rule: hold QQQ in month t+1 if the close at the end of month t is above the average of the last 10 month-end closes. No other lengths were tested.
Sources
- Faber, M. T. (2007). A quantitative approach to tactical asset allocation. Journal of Wealth Management, 9(4), 69–79. Working paper, updated February 2013: SSRN 962461
- Park, C.-H., & Irwin, S. H. (2007). What do we know about the profitability of technical analysis? Journal of Economic Surveys, 21(4), 786–826. IDEAS/RePEc
- Alpha Vantage. TIME_SERIES_MONTHLY, DIVIDENDS and TREASURY_YIELD. Data retrieved October 2026.
FAQ
What is the 10-month moving average rule? Hold an asset when its month-end price is above the average of its last 10 month-end prices. Otherwise, hold cash.
Does the 10-month moving average work on the Nasdaq? From 2000 to 2026 it improved returns and cut the worst drawdown from −75% to −25%. Since 2013, it's trailed buy-and-hold.
Is the 10-month moving average the same as the 200-day? They're close: ten months is about 210 trading days. The 200-day uses daily prices and changes every day.
How often does the rule trade? On QQQ, 47 switches in 26 years, under two a year.
Is a moving average a good trading signal for NQ? At the monthly scale, it's a regime filter. It's not built for intraday entries.
Why did the rule lag buy-and-hold after 2013? There was no crash deep enough to avoid, and several corrections caused whipsaws, like six switches in 2016.
Research and education only. Not financial advice. Past results do not predict future returns. A backtest is not real performance.
How this article was made: drafted with AI assistance from public data and the sources listed above; facts and figures checked by Gabriel Décarie and Karel Panneton.
The week just closed, summarized, and the economic calendar for the week ahead.