Rank every stock in the Nifty 500 by trailing 12-month return and take the top 30. In a year when metals ran, you will hold mostly metals. The screen said momentum and the portfolio said commodities.
Sector-neutral construction breaks that link. Instead of computing one cross-sectional score across the whole universe, you group stocks by sector and standardise within each group. A bank is scored against banks, a pharma company against pharma companies. The result is a score that answers “how strong is this stock relative to its peers” rather than “how strong is this sector right now”.
There are two places to apply it. Sector-neutral scoring changes the ranking, so selection is made on relative-to-peer strength. Sector-neutral weighting fixes each sector’s total portfolio weight, usually at the benchmark weight or at an equal share across sectors, and then distributes within the sector by score.
What changes
| Approach | Selection | Typical portfolio |
|---|---|---|
| Universe-wide score | Best absolute scores | Concentrated in whichever sectors are trending |
| Sector-neutral score | Best score within each sector | Spread across sectors, stock-level bets only |
| Sector-neutral weight | Any | Sector weights fixed, active risk is stock selection |
Why it matters in the Indian market
The Nifty 50 is not sector-balanced. Financial services is by a wide margin the largest block in the index, and a naive value screen on Indian equities has historically leaned heavily into banks and public-sector names because those are where the low multiples sit. If you screen for cheapness and end up 60 percent in financials, your return will be explained by what happened to Indian banks that year, whatever your rule was named.
A maximum group weight constraint achieves something similar from the other direction: leave the scoring alone, and cap the total weight any one sector can reach.
The case against
Sector neutrality can remove the bet you actually wanted. If your thesis is that low-volatility stocks outperform, part of that effect is structural: FMCG and utilities are genuinely calmer businesses than metals and real estate. Neutralise sectors and you have stripped out a real component of the anomaly, leaving only the within-sector residual, which is smaller and noisier.
Sector definitions are also a modelling choice, not a fact. Whether a lending-heavy conglomerate sits in financials or industrials changes its peer group and therefore its score. Sectors with few members are the sharpest edge case: standardising within a group of four stocks produces a mean and a standard deviation you should not trust, and the top-ranked name in that group gets selected on almost no information.
Decide first whether your edge is a stock-picking edge or a sector-timing edge. Neutralise only the one you are not trying to make.