Crypto's momentum formula loses money on Indian stocks. A six-month version doesn't.
A residual-momentum strategy has held up in crypto for years: rank instruments by how far they have outrun the broad market, buy the leaders, short the laggards, rebalance daily. We pointed the same rules at ten years of Indian equity data. They did not travel. Applied unchanged to Nifty 500 stocks, the Sharpe ratio came out at 1.1 in the negative, against roughly 1.0 positive out of sample in crypto. Sharpe ratio here means return measured against how violently that return swung about; below zero, the strategy was not merely noisy, it was pointed the wrong way.
The reason is structural. Short-horizon price behaviour runs in opposite directions in the two markets. Crypto keeps going for days at a time. Indian equities snap back over a week or a month and only begin to trend at horizons of 6 to 12 months. Three frictions deepened it: once market beta is stripped out, the dispersion left between individual stocks is thin, the factor is crowded and heavily arbitraged, and equity costs bite harder.
If shares reverse over days, trade the reversal instead. Every version of that flip lost money after costs, landing between 0.6 and 1.5 in the negative on Sharpe; among liquid names the short-horizon bounce has already been arbitraged away. Standard refinements fared no better. Neutralising sector exposure, weighting by inverse volatility, scaling to a volatility target: all of them hurt. The premium appears to sit in precisely the dimensions those adjustments strip out.
What finally worked was slower and less clever. Momentum measured against the market over six months rather than a week, with the most recent week ignored. Weekly rather than daily rebalancing, equal weights, a universe cut to the hundred most liquid names. A deliberate net-long tilt of 0.7 long against 0.3 short instead of a balanced book. And a switch entirely to cash whenever the index sat below its 200-day average price. Net of 16 basis points of trading cost, that version produced a Sharpe of 1.02 across the full period and 1.32 out of sample, compounded at 12% a year, was positive in 8 of 10 calendar years, and had a worst year of 4% down. It still cleared its costs at 30 basis points.
Three changes did the work. Lengthening the horizon accounted for the change of sign. The net-long tilt gave the largest single lift, moving the strategy from six positive years out of ten to eight. The move-to-cash rule addressed momentum's classic failure mode, the sudden reversal: in 2025 it converted a 10% loss into a 3% one.
Then comes a catch no tuning removes. That short sleeve has to be held for weeks, which in India means single-stock futures; a plain cash account cannot do it, because a cash short must be closed out the same session. So the book was rebuilt without shorts: long only, always invested, the same six-month ranking refreshed weekly, holding the top fifth of the list, around 20 stocks, taken to delivery. Positive in 8 of 10 years, Sharpe 0.85 in sample and 1.09 out of it, compounding near 20% a year, worst year 12% down in the 2025 selloff.
Where the costs land is the genuinely unusual part. Realistic delivery charges here come to roughly 24 basis points, counting securities transaction tax at 0.1% on each side plus stamp duty and depository fees, and subtracting them barely moves the outcome. The book holds up at 40 to 60 basis points, because it turns over slowly: about 6 one-way fills a week, 28 a month, 330 a year. That is the reverse of the intraday setups we have tested, where the fill price is the whole problem.
Compounding near 20% is not the impressive part, and calling it skill would be dishonest. Most of it is exposure to a market that rose over the decade. The rest is inflated by survivorship bias: the test ran on the companies in the index today, which by construction leaves out those that shrank out of it or failed. Crash protection bolted onto the long-only book — per-stock trend filters, an index cash signal, breadth gates — only whipsawed. The short leg had been the only real cushion.
One further overlay pushed risk-adjusted returns above one, and it produced the study's most instructive failure. Split the index's ten-year drift by session and it separates into two very different halves: 15.7% a year lost while the market is open, against 28.2% a year gained between the close and the next open. Holding the momentum longs overnight while shorting the index during the session lifted the long-only book to a Sharpe of 1.42 at a full hedge and 1.15 at half, positive in 9 years out of 10. Then we ran the attribution. A plain equal-weighted long book with the same intraday short, carrying no momentum ranking at all, scored 2.24 — higher. The overnight gap was the whole source of the gain, and momentum, if anything, dragged on it.
Several things argue for reading even the modest hedged number as optimistic. The intraday short enhances returns; it does not protect against a crash, and it made the bad year worse, deepening that 12% fall to 19% at a full hedge. The session split is measured against exact opening and closing marks nobody actually transacts at, and roughly 250 index shorts a year each carry real slippage: the full hedge stops working at 8 basis points a day, the half hedge survives to about 5.
Honest summary: this family of momentum can work across years on Indian equities, but only once rebuilt into something its crypto version would not recognise. The variant an ordinary delivery account could have run scored 0.85, its decade of compounding flattered by direction and by survivorship. Everything above that came from a separate effect — one widely documented, prone to crowding, and partly just payment for overnight gap risk a long book already carries. If live slippage reaches 8 basis points, the above-one figure reverts to roughly 0.8. All of it describes what past data did, nothing more.
More money stories
- Two of eight sessions made the entire profit — and the short side lost money
- Low-volatility stocks, not momentum, produced a book positive in every completed year
- We tried to make a momentum strategy win every year. The momentum had to go.
- Fading morning gap-ups paid in liquid stocks and lost in illiquid ones
- Widening the stock pool, not the signal, gave this hedged low-volatility test a positive decade
- A short-volatility test on Nifty returned 11.3% a year, inside a 15% drawdown