A factor is a stock attribute you can measure for every name in a universe and rank on. Trailing 12-month return is a factor. Return on equity is a factor. Realised volatility over the last 90 days is a factor. The claim behind factor investing is that some of these attributes have historically separated future winners from losers with enough consistency to be worth building rules around.
India has a live market for this. Smart-beta and factor AUM grew from roughly ₹290 crore in 2020 to about ₹46,000 crore by end-2025 across 70-plus factor ETFs and index funds. NSE publishes the underlying indices, including the Nifty 200 Momentum 30 and the Nifty 100 Quality 30, which select on a factor score and then weight by that score subject to a cap so the highest-scoring name cannot dominate the index.
Common factors
| Factor | Typical measure | Thesis |
|---|---|---|
| Momentum | 12-month return, skipping the last month | Recent relative winners continue |
| Value | Earnings yield, book to price | Cheap assets are underpriced |
| Quality | Return on equity, low debt, stable earnings | Durable businesses compound |
| Low volatility | Trailing standard deviation, beta | Calm stocks have delivered better risk-adjusted returns |
| Size | Market capitalisation | Smaller companies carry a risk premium |
Combining factors into one score
A single factor is a single bet with long stretches of underperformance. Combining several is standard practice, and the mechanics matter. Raw factor values live on incompatible scales: a return on equity of 22 and a 12-month return of 0.34 cannot be added. Each factor is standardised first, using a z-score, a rank, a percentile or a min-max rescale, then the standardised values are combined with weights into one composite score.
Two options change the character of that score. Winsorisation clips extreme values before standardising so one outlier does not dominate. Group-by scoring standardises within each sector rather than across the whole universe, which removes the sector tilt that most raw factor screens carry.
Where factors disappoint
Every documented factor has multi-year drawdowns. Value underperformed globally for most of the decade to 2020, and an investor who abandoned it in year six locked in the loss. If you cannot hold a factor through five bad years, the historical premium is not available to you.
Crowding is the second issue. As factor AUM grows, the same ranked names get bought by the same rules on the same reconstitution dates, which compresses the spread and can turn rebalance days into crowded exits.
Then there is the research problem. Many published factors were found by searching the same historical data many times, and a result that survives 300 tests on one dataset is not evidence. Prefer factors with an economic story you can state in one sentence, and test them out of sample.