Every stock-selection screen has the same blind spot. Ask it for the thirty best-ranked names and it returns thirty names — in a bull market, in a correction, in a crash. “Best of a falling universe” is still a fully invested equity position, and no amount of selection skill compensates for holding 100% equity into a 50% index decline.
Market-level timing addresses a different question from stock selection: not which stocks, but whether to be in stocks at all.
The rule and the evidence
Mebane Faber’s 2007 paper, one of the most downloaded on SSRN, tested a rule of near-embarrassing simplicity: hold the asset when its price is above its 10-month moving average, hold cash when it is below. Check once a month.
Applied to the S&P 500 from 1901, the rule produced returns broadly comparable to buy-and-hold while cutting volatility substantially and, most notably, reducing maximum drawdown from around 83% to roughly 50%. It sidestepped the worst of 1929–1932, 1973–1974, 2000–2002 and 2008. The same rule applied across foreign equities, bonds, commodities and REITs produced similar risk reduction.
The 200-day moving average is the daily analogue of the 10-month, and behaves nearly identically.
Read the result correctly
The headline that circulates is “beat the market with less risk.” That is not what the paper shows, and the distinction matters for whether you should use it.
Returns were comparable, not better. The improvement is concentrated entirely in the risk metrics — volatility and, above all, maximum drawdown. This is a drawdown trade.
Which is not a small thing. Drawdown, not average return, is what determines whether a real person is still holding a strategy at the bottom. A portfolio that falls 50% needs a 100% gain to recover; one that falls 25% needs 33%. And the investor who capitulates at the bottom of the 50% decline realises the loss and misses the recovery, which is the outcome the arithmetic of average returns never captures.
The cost is real and you will feel it
The rule fails in a specific, predictable way: whipsaw.
In a choppy, directionless market the index crosses its 200-day average repeatedly. Each crossing triggers an exit near a local low and a re-entry at a higher price. Do that four times in eighteen months and you have paid four round trips of costs and taxes to end up behind buy-and-hold. This is not a rare edge case — sideways markets are common, and the rule underperforms in every one of them.
There is also a structural cost: the strategy is always late at both ends. It never sells at the top or buys at the bottom, because it waits for confirmation. In exchange for never being early, it gives up the first leg of every recovery.
Anyone deploying this should expect to underperform a buy-and-hold benchmark in most individual years, and to be very glad of it in the one year out of ten that matters. If that pattern would cause you to abandon the rule in year three, the rule is not for you — an abandoned timing strategy is worse than no timing strategy, because you take all the whipsaw costs and none of the crash protection.
Gating a screen rather than an index
Faber’s construction times an index directly. Applying the same logic to a stock-selection strategy raises a design question: what does “risk off” mean for a screen?
Two answers, and only one is honest.
Keep the screen running but hold nothing. The gate evaluates the index condition first; if it fails, the reconstitution produces an empty portfolio and existing holdings are sold. The strategy sits in cash until the index recovers.
Keep the screen running and hold anyway. This is not timing. It is a comment.
The first is what the template does, via a rule that gates the whole reconstitution on one instrument’s condition — here the Nifty 50 against its own 200-day SMA — and exits the portfolio when the condition fails.
Note that the gated instrument is deliberately not one of the holdings. It is the market. A stock-level trend filter is a different strategy with different properties; this one is asking a single question about the environment, and applying the answer to the whole book.
A second, independent brake
An index-level filter watches the market. It does not watch your portfolio, and those can diverge — a concentrated momentum book can be in a 30% drawdown while the index is comfortably above its average.
A portfolio circuit breaker closes that gap by monitoring the strategy’s own equity curve and halting when drawdown from peak exceeds a threshold. The two controls are independent, which is the point: one responds to the market, the other to what the strategy has actually done to your capital.
Setting the threshold is a judgment call. Too tight and normal volatility trips it, locking in losses at the worst moment. Too loose and it never fires. For a monthly-rebalanced equity strategy, something in the 20–30% range is far enough out that only a genuine failure reaches it.
What this does not fix
The 200-day rule protects against slow declines — the multi-month grind where the index breaks its average early and keeps falling. It cannot protect against a gap. If the market falls 12% overnight on an exogenous shock, the rule has not had a chance to fire and you are fully invested through it.
March 2020 is instructive on both counts. The initial collapse was too fast for a monthly-checked rule to avoid much of; the subsequent recovery was fast enough that the rule kept you out of a substantial part of it. That episode was close to the worst case for this family of strategies, and it is worth including in any window you test over rather than starting the backtest conveniently after it.
Try it
The Trend-Gated Momentum template runs the same 12-1 momentum screen as the ungated version, with two additions: reconstitution only proceeds while the Nifty 50 is above its 200-day SMA, and a 25% portfolio drawdown halts the strategy.
Run both over an identical window and compare three numbers: total return, maximum drawdown, and number of rebalances. The gated version will usually give up some return, cut the drawdown meaningfully, and trade more. Whether that is a good trade is a question about you, not about the backtest.
Further reading
- Volatility targeting: sizing is a more reliable lever than timing is the case for adjusting exposure continuously instead of switching it off
- Commodity trend following on MCX applies the same filter where the mechanism behind the trend is physical rather than behavioural
- Cross-asset momentum: the version that held up out of sample for a strategy that uses this rule as one leg rather than as the whole trade
Glossary: simple moving average, max drawdown, drawdown duration, market breadth.