Dual momentum tripled the Nifty's risk-adjusted return over 19 years, and halved the drawdown
Most of what fund houses run in-house is useless to an individual investor: too fast, too capital-hungry, or dependent on execution nobody outside a trading desk can get. So Strota asked a narrower question. Strip the list down to systematic algorithms a person could actually operate, and does any of them still work on the Nifty? One did. A monthly rule known as dual momentum produced a Sharpe ratio of 0.76 across the test window, against 0.25 for simply buying the index and sitting on it.
Sharpe ratio is the standard way of asking how much return a strategy delivered per unit of the wobble it put you through. Higher is better, and a gap that size is not a rounding artefact. The same rule also cut the damage sharply. Buying and holding the index gave up 60% from peak to trough at the worst point in the period, nearly all of it during the 2008 collapse. The dual-momentum rule's deepest fall was 22%. Compounded growth landed at 12.3% a year, against 9.5% for the index itself.
Two crude checks, run once a month, are all the rule consists of. The first is absolute: has the Nifty itself gone up over its trailing window? If not, the rule leaves equity entirely and sits in gold or cash. The second is relative: between the index and gold priced in rupees, hold whichever has the stronger trailing return. Nothing is forecast. It reads what has already happened and follows it.
Why that works is not mysterious. The absolute check is what kept the strategy out of the worst equity episodes in the sample, 2008, 2020 and 2022, because a falling index fails its own test and the rule steps aside. Gold does the other half of the job, since it has tended to hold up when Indian equity is being sold hard.
Two rival candidates were run on the same data, and neither justified itself. A long-or-flat trend filter, in the market while price sits above its long-run average and out when it drops below, trimmed the worst drawdown to 20%, but its Sharpe of 0.23 was no improvement on doing nothing, and compounded growth slipped to 8.5%. Volatility targeting, which shrinks position size when the market turns jumpy, matched plain holding at 0.25 while cutting the worst fall to 36%. Both bought comfort. Neither bought edge.
The study ran across 19 years, from 2007 to 2026, deliberately including the 2008 crash so the sample was not a bull market with the disaster edited out. Index returns came from Nifty history; gold was expressed in rupee terms by converting a dollar gold series at the prevailing exchange rate. Every headline figure is net of the running costs of the exchange-traded funds involved, and each switch was charged ten basis points.
Cost treatment matters less here than almost anywhere else, because the rule barely trades. It fired between 1.5 and 2.7 switches a year. When a strategy turns over roughly twice annually, fees are a footnote, which is the reverse of most intraday ideas, where costs decide the outcome before the signal gets a say.
A single flattering parameter is the usual way a backtest lies, so the trailing window was swept rather than picked. Every setting tried from six months out to fifteen produced a Sharpe between 0.75 and 0.83, compounded growth of 12% to 13.6%, and a worst drawdown that never ran past 25%. The result does not balance on one lucky choice.
Splitting the record by era is the harsher test, and it is where most strategies come apart: they bulge in the early years, where the rule was effectively designed, and go flat afterwards. This one did not. Measured from 2007 the Sharpe was 0.75. From 2013 it was 0.73. From 2018, the out-of-sample stretch, 0.77. Near-identical readings across three overlapping windows is the fingerprint of something real rather than something fitted.
Execution is the hurdle a fund-house method usually fails, and this one clears it. Three liquid, widely traded exchange-traded products cover the whole thing, one tracking the index, one tracking gold, one parking cash, with a single decision taken at the month's closing price.
The limits deserve stating plainly. Gold in the study was a proxy, built from a dollar gold price and the rupee exchange rate. A real Indian gold fund carries tracking error and expenses of around 0.5% that the proxy never pays. That is a genuine drag on the figures above, not a decimal quibble.
Nor is 12% a year spectacular by Indian equity standards. The finding is not about raw return; it is about return per unit of pain, and about a peak-to-trough fall roughly half as deep. Read as a growth number, it has been read wrong. The rule also has a known failure mode: a directionless, choppy market can flip it back and forth in small unrewarding increments, and while the switch cost is modelled, the tedium of sitting through that is not.
None of this describes something on offer now. It describes what a fixed set of mechanical rules would have produced across a specific 19-year slice of Indian market history, measured after costs, with the weak points named. What makes it worth reporting is not that a rule made money once, which backtests manage all the time, but that it kept behaving the same way in every period it was cut into, using an approach slow and capacious enough that being widely known does not obviously kill it.
More money stories
- Boring stocks held up better than winners in our re-test of Indian factor books
- Five ways to build a 0.5% intraday trade, and the cost wall that stopped all of them
- Strota backtested its own Darvas box screen: no edge once you pay the fees
- Four attempts to beat intraday trading costs failed -- the barrier is the fill, not the signal
- A rare positive in our backtests: deep gap-down buying cleared the cost wall, at retail scale only
- Gold crosses Rs 1.5 lakh per 10 grams, and silver is the tighter story