Skip to content
StockMarketAgent
Methodology · Sector rotation

Sector rotation matrix

A heatmap of sector-level relative returns across multiple lookback windows used to characterize whether index leadership is rotating early-cycle, late-cycle, or defensive.

Specification

Sector rotation matrix — operational spec

Visualization aid that corroborates the regime read by showing which sector cohort is doing the work in the index.

Inputs
  • Daily total-return series for the 11 GICS sectors (sector ETFs as a proxy)
  • Daily total-return series for the broad benchmark (S&P 500)
  • Three lookback windows: 1 month, 3 months, 12 months (configurable)
  • Neutral threshold (default 0.5 percentage points absolute)
Computation
  • For each (sector, window) pair, compute sector total return minus benchmark total return over the window.
  • Cells are tinted bull, bear, or neutral by the sign and magnitude of the spread; absolute values below the neutral threshold render flat.
  • No smoothing is applied — the matrix shows the raw relative return so noise is visible rather than hidden.
  • Sectors are ordered by their 12-month spread descending, so the strongest cohort is at the top.
Outputs
  • 11 x 3 matrix of relative-return spreads with bull / bear / neutral tinting.
  • Sector ordering by 12-month relative strength.
  • Implied leadership profile (early-cycle / late-cycle / defensive) inferred from which cohort is dominating each window.
Limitations
  • Sector momentum is a notoriously poor cross-sectional return predictor on its own. The matrix is descriptive, not predictive.
  • ETF proxies for the GICS sectors carry tracking error and a small management-fee drag relative to the underlying index.
  • The cycle-mapping inference is heuristic. Atypical regimes (e.g. concentrated mega-cap leadership) can defeat the early- vs late-cycle pattern.

Frequently asked

Why three windows?
Short, medium, and long lookbacks isolate different signals. A sector that leads on 1M but not 3M / 12M is in a tactical bounce. A sector that leads on 3M / 12M but lags on 1M is a structural leader losing momentum. Three windows is the minimum that disambiguates.
Can the matrix predict the next leading sector?
No. The cross-sectional return predictability of sector momentum is weak and unstable across regimes. The matrix is a descriptive aid that helps the reader corroborate the regime read; it is not a sector-rotation strategy.
Why ETF proxies and not the underlying index?
ETFs deliver point-in-time tradable returns including dividends and rebalancing, which is what an investor actually receives. The tracking error against the underlying GICS sector index is small and well-documented; the ETFs are the more honest data source.
See the full methodology hub for the rest of the model registry, or open the glossary entry for the headline definition.