Momentum is one of the most well-documented anomalies in financial research. The strategy — buy stocks that have outperformed over the past 12 months (excluding the most recent month) and avoid or short underperformers — has been documented across 40+ markets globally. India is no exception.
The NSE Momentum Premium: 2015–2025
Backtesting a long-only momentum strategy on the Nifty 500 universe (top-quintile 12-1 month returns, equal-weighted, monthly rebalancing) from January 2015 to December 2025 yields an annualised return of 19.2% versus the Nifty 500 TRI's 14.8% — a momentum premium of approximately 4.4% per annum before transaction costs. After realistic trading costs (0.15% per trade, brokerage + STT), the net premium narrows to 2.8%.
Sector Concentration Risk
Momentum portfolios on NSE exhibit dangerous sector concentration during trend regimes. In 2021, the momentum portfolio was 60%+ weighted in real estate and metals. In 2023, it rotated entirely to capital goods and defence. This concentration means momentum drawdowns are not independent of sector crashes — a sudden regulatory shock to a dominant sector produces momentum crashes of 25–35%.
Managing Momentum Exposure
Practitioners in Indian markets have found that combining momentum with quality filters significantly reduces crash risk:
- Quality screen: Restrict the momentum universe to stocks with ROE > 15% and debt/equity < 1. This eliminates leveraged cyclicals that dominate momentum portfolios at cycle tops.
- Volatility adjustment: Scale position sizes by inverse volatility — high-momentum/low-volatility stocks like Bajaj Finance or Asian Paints receive larger allocations than high-momentum/high-volatility smallcaps.
- Regime filter: Momentum underperforms in mean-reverting regimes (high VIX, post-crash recoveries). Spectrum's Regime Detection module flags Bear and Sideways regimes where momentum exposure should be reduced.
The bottom line: momentum works on NSE, but naive implementation destroys returns through crashes and taxes. A disciplined, factor-aware approach is essential.