Trend rules halve the worst falls in Indian stocks, and cost you for it

The idea is old and simple. Hold the market while its price is above some average of its recent past, and move to cash when it falls below. Mebane Faber’s 2007 paper made the ten-month version famous: on a century of US data it kept most of the market’s return and avoided most of its crashes.

We tested five such rules on seven Indian assets: the Nifty 50, Nifty 500, Next 50, Midcap 150, Smallcap 250, gold and a five-year government bond index. The rules and their parameters came from the published literature and were written down on 1 October 2026, before any Indian result was computed. That is 35 combinations of rule and asset, and all 35 are reported.

What the rules deliver

They deliver smaller falls. On the Nifty 50, buying and holding since 2000 meant sitting through a fall of 59.5% in 2008. Every trend rule cut that to between 27% and 39%, because each was mostly in cash for the worst of the crash.

On the Nifty 50, the 200-day rule ended behind buying and holding

Total return, before tax, after trading and fund costs. The rule switches to cash (earning T-bill yields) when the index closes below its 200-session average.

200-day average ruleBuy and hold

Loading chart…

The price was paid in return. The rules’ annual returns on the Nifty 50 were 0.5 to 4 points below buy-and-hold’s 12.3%. Measured per unit of risk, the best of them edged ahead: Sharpe ratios ranged from 0.21 to 0.41, against 0.35 for simply holding.The Sharpe ratio here is the annual return over cash divided by the annual volatility of returns. Cash earns the 91-day T-bill yield.

Was the edge luck?

When 35 strategies are tried, the best one will look good by chance. The deflated Sharpe ratio asks how likely it is that a strategy’s true Sharpe beats the best of that many worthless ones:

$$\text{DSR} = \Phi\!\left( \frac{(\widehat{SR} - SR_0)\sqrt{T-1}}{\sqrt{1 - \hat{\gamma}_3\,\widehat{SR} + \tfrac{\hat{\gamma}_4 - 1}{4}\,\widehat{SR}^{2}}} \right)$$

where $SR_0$ is the Sharpe ratio the best of 35 null strategies would be expected to reach, $T$ is the number of observations, and $\hat{\gamma}_3$ and $\hat{\gamma}_4$ are the skewness and kurtosis of returns. The usual bar is 0.95.

Against buy-and-hold, no large-cap rule came close: the deflated value for the edge over holding the Nifty 50 was 0.03 or less for every rule.

The mid and small caps are where the rules looked best. The 200-day rule earned 18.1% a year on the Midcap 150 against 15.4% for holding, with a worst fall of 32% instead of 73%.

On midcaps, the 200-day rule ended ahead before tax

Nifty Midcap 150 total return, before tax, after trading and fund costs.

200-day average ruleBuy and hold

Loading chart…

Even this result does not survive deflation. Its deflated value against buy-and-hold is 0.16, and the Smallcap 250’s, where the rule earned 18.3% against 13.8%, is 0.29. Both are far from 0.95.

Then comes tax

Each exit from a trend rule realises a gain, often a short-term one, and the cash leg’s interest is taxed at the slab rate. Rerunning every rule with Indian capital-gains rules as they stood on each date changes the picture.¹

After tax, every trend rule trailed buy-and-hold on every large-cap index, by 1.9 to 6.1 points a year. Holding the Nifty 50 returned 11.98% a year after tax, against 12.31% before. The ten-month rule returned 9.58% and the 200-day rule 8.28%. On midcaps the 200-day rule’s edge of 2.69 points before tax became −0.04 after.

Method

Rules. sma10m: long when the month-end level is above the average of the last ten month-ends (Faber). sma200d: long when the close is above its 200-session average. tsmom12: long when the trailing 12-month return beats cash. blend: exposure equal to the share of 1, 3, 6 and 12-month excess returns that are positive. breakout: long on a 252-session high, out on a 126-session low.

Timing. Signals use the close of day $t$; the position changes at the close of day $t+1$, so no rule sees a price before it could have traded on it.

Costs. 0.10% of value traded each way, plus fund running costs while invested: 0.15% a year for equity, 0.50% for gold, 0.20% for government bonds and for cash.

Data. NSE total-return indices, with bad prints repaired against the price index. Gold is the domestic rupee price including import duty.

Robustness

Halves of the sample. Results are reported separately for the first and second halves of each history and from 2012, in the results bundle.

The weakest rules. The 12-month momentum rule and the breakout rule did worst across assets.

Gold and bonds. Trend rules lowered or matched returns, with no benefit, and lost 0.9 to 6.4 points a year after tax.

Short history. India has about 25 years of data for most indices and about 20 for mid and small caps. That is few independent trend cycles, so the uncertainty around every number here is wide.

The verdict

The part of Faber’s claim about drawdowns travels to India intact. The part about keeping the return does not survive a correction for the number of rules tried, and for a taxable Indian investor the rules historically bought smaller falls at a large cost. Inside a tax-free wrapper such as a fund or the NPS, the pre-tax numbers are the relevant ones.

Footnotes

1. The tax model uses first-in, first-out lots and the rates in force on each sale date, including the 2018 grandfathering of equity gains. It ignores cess, the annual exemption and loss set-off, and sells everything at the end. ↩