Loading Nifty 50, Nifty Bank, and Sensex…
Cross-sectional research

Sector & Size Neutral Factors

The same Factor Lab scores, with sector and company size regressed out — so a high score is not quietly a small-cap or single-sector bet.

Factor Lab
Ctrl-click or drag to select several; they are averaged into one raw score.
Rows returned, not listings scored.
What this page is for
What "neutralized" means

Rank any exchange on a value or quality score and the top of the list fills up with the same kinds of company — one or two sectors, and usually the smaller names. That is the sector and the size talking, not the factor.

This page fits each raw score to sector membership and company size (log market cap) across the whole cross-section, then keeps only the residual — the part of the score those two things do not explain. A high neutral score means the listing scores well against companies like it, not merely that it is small or in a cheap sector.

What people use it for
  • Stopping a screen from becoming one bet. A raw factor list is often a small-cap list wearing a factor label.
  • Comparing like with like. A large bank and a small one score on different scales; neutralizing puts them on one.
  • Measuring the bias itself. The sector-shift and size-shift columns say exactly how much of each raw score was sector and size.
  • Checking a conviction. If a name only ranks well before neutralizing, what you liked was its cohort.
What it cannot tell you

This re-expresses the same evidence — it adds no data and makes no claim about future returns. Removing sector and size does not make a factor predictive, and a residual ranking has its own habits: it can favour listings with unusual accounting or thin cohorts, where "unlike its peers" and "wrongly measured" look identical. Read it next to the raw score, never instead of it.

What each column means
Neutral score — the residual after sector and size are regressed out, in standard-deviation units — roughly, how many typical deviations the listing sits above or below what its sector and size alone would predict. 0 is exactly as expected for its cohort; 2 is far above it. This is what the table ranks on.
Raw score — the original Factor Lab percentile, 0–100, before anything is removed. A high raw score with a low neutral score means the cohort did the work.
Sector shift — how much the sector adjustment moved this listing's score. Large negative values mean the raw score was largely a sector effect — its whole sector scored high, so scoring high there was not distinctive.
Size shift — the same for company size. Small values mean size had little to do with the raw score for this listing.
Sector cohort — how many listings shared this sector in the regression. A residual measured against three peers is a much weaker statement than one measured against forty — treat tiny cohorts with suspicion.
Size coefficient — the exchange-wide relationship between company size and the raw score. Negative means the raw factor scored smaller companies higher across the board; near zero means size barely mattered and neutralizing changed little.
Scored universe — how many listings had enough persisted data to be scored at all, out of the full exchange. Everything else is excluded rather than estimated, so the ranking is over what could actually be measured.
Note — why a row could not be neutralized — usually a missing input or a sector cohort too small to regress against. Those rows stay visible and unranked instead of quietly disappearing.
Choose an exchange and factors to rank on the part of each score its sector and size do not explain.

A neutral score re-expresses the same evidence. It carries no claim about future returns and is not a recommendation.