There are several meanings of neutral
Sector neutrality can mean ranking stocks against sector peers, matching a benchmark's sector weights, or making long and short sector exposures offset. These are different operations. A portfolio can use within-sector scores and still overweight a sector if it selects more names there. It can also match sector weights while holding unusually volatile or highly leveraged companies within each sector.
MSCI's sector-neutral quality description provides a concrete example of evaluating companies relative to peers within the same sector. This does not make every sector-relative model a replication of that index. An independent specification must identify its sector taxonomy, historical classifications, benchmark, weighting rules and tolerance for deviations. The word neutral should always be followed by the exposure and reference being neutralized.
Match sectors and still change beta
Assume a fictional benchmark has 30% in technology and 70% in industrials. Its sector-sleeve betas are 1.1 and 1.0, giving a weighted beta of 1.03. A portfolio matches those sector weights but selects stocks whose sleeve betas are 1.6 and 0.9. Its estimated beta is 1.11, leaving 0.08 of active beta despite zero active sector weight.
The arithmetic assumes betas are measured against the same market, on compatible observations and with fixed weights. It is not a forecast of a market move. It simply shows why matching two sector totals does not constrain the characteristics inside those totals. Size, valuation, country, currency and business concentration may also remain different.
Choose what the experiment is allowed to express
If the hypothesis concerns stock selection within industries, constraining broad sector weights can help isolate it. If the hypothesis deliberately includes sector rotation, enforcing neutrality removes part of the intended signal. Decide this before reviewing performance. Comparing an unconstrained winner with a constrained loser after the fact does not reveal which economic hypothesis was originally tested.
Sector definitions themselves are dated data. A company can change classification, and applying its current sector to all historical records introduces information and comparability problems. Retain the classification effective date and source version. For very small peer groups, within-sector normalization may be unstable; establish a minimum group size or broader fallback classification rather than inventing a score from two observations.
Report residual exposures and constraint costs
Show benchmark weights, portfolio weights and active differences together. For long-short portfolios, include both gross and net sector exposures: a net zero can conceal a large matched book. Track issuer overlap, beta estimates and factor sensitivities separately. If a sector cap forces low-ranked stocks into the portfolio, report the resulting signal dilution and incremental trading as part of the design.
Neutrality reduces a specified exposure only under its measurement assumptions. It does not guarantee diversification during a shock, eliminate model error or prove that residual returns are skill. A credible evaluation compares predeclared constrained and unconstrained variants on identical data and timing, including costs. The example provides no historical sector returns and makes no claim that the higher- or lower-beta allocation is superior.
Inspect the example
| Illustrative sleeve | Benchmark weight | Portfolio weight | Benchmark beta | Portfolio beta |
|---|---|---|---|---|
| Technology | 30% | 30% | 1.10 | 1.60 |
| Industrials | 70% | 70% | 1.00 | 0.90 |
| Weighted total | 100% | 100% | 1.03 | 1.11 |
Sources & reproduction
Primary references support the definitions and method. Numerical examples are illustrative unless a source record is explicitly identified.