US Sector Rotation Monitor
Systematic Sector & Industry Leadership Intelligence. V2.12 preserves the production sector score and adds the independently validated Value Cycle Overlay as a transparent live-shadow research layer.
1. Purpose and Scope
The monitor ranks eleven US sector ETFs and a separate universe of industry ETFs. It combines relative momentum, absolute trend, market internals, macro context and parent-sector confirmation. The model is designed for cross-sectional research and monitoring; it is not a personalized portfolio, probability forecast or execution system.
2. Sector Universe
The core universe comprises the eleven US sector ETFs: XLK, XLC, XLY, XLP, XLE, XLF, XLV, XLI, XLB, XLRE and XLU. SPY is the primary market benchmark. Additional factor and breadth ETFs are used for internal confirmation.
3. Sector Score Architecture
The sector score is a robust weighted average. Missing non-critical components are ignored and remaining weights are renormalized. A minimum available-weight threshold is required before a score is published.
| Component | Raw coefficient | Treatment |
|---|---|---|
| rel_mom_63_z | 0.28 | Renormalized over available signals |
| rel_mom_21_z | 0.16 | Renormalized over available signals |
| rel_mom_126_z | 0.12 | Renormalized over available signals |
| trend_distance_z | 0.14 | Renormalized over available signals |
| abs_mom_63_z | 0.10 | Renormalized over available signals |
| abs_mom_126_z | 0.06 | Renormalized over available signals |
| trend_health_score | 0.08 | Renormalized over available signals |
| volume_impulse_z | 0.04 | Renormalized over available signals |
| sector_macro_tailwind | 0.12 | Renormalized over available signals |
4. Score Scale and States
The raw score is clipped and transformed to a 0–100 scale, then smoothed with a five-observation exponential moving average. State bands are:
| State | 0–100 range |
|---|---|
| Leading | ≥ 71.25 |
| Improving | 57.50 to 71.25 |
| Neutral | 42.50 to 57.50 |
| Weakening | 28.75 to 42.50 |
| Lagging | < 28.75 |
5. Industry Drilldown
Industry ETFs are scored independently from the sector core. Each industry is evaluated versus SPY and versus its parent-sector ETF. This layer is used to distinguish confirmed, narrow, hidden and divergent leadership.
| Industry component | Weight |
|---|---|
| Market relative momentum 3M | 0.24 |
| Market relative momentum 1M | 0.10 |
| Market relative momentum 6M | 0.08 |
| Parent relative momentum 3M | 0.25 |
| Parent relative momentum 1M | 0.08 |
| Absolute momentum 3M | 0.10 |
| Trend distance | 0.07 |
| Trend health | 0.06 |
| Volume impulse | 0.02 |
6. Value Cycle Overlay
Value Cycle is an independent long-only allocation overlay; it does not change the production Sector Rotation Score. The factor input is the official daily Fama-French HML series from the Kenneth French Data Library. Sector sensitivity is estimated from 126-session rolling betas of sector returns relative to SPY. The HML impulse is measured over 126 sessions, standardized by 252-session volatility and cross-sectionally ranked after multiplication by sector beta.
The complete signal uses a fixed 42-trading-session information lag, a five-observation exponential smoother and a confidence gate at an absolute HML strength of 1. Monthly signals enter at the next trading close and are averaged across twelve cohorts. When active, the model transfers up to 20% from low-ranked to high-ranked sectors around a dynamic equal-weight sector baseline. The historical series includes 25 basis points of transaction costs under the research turnover convention.
The locked backtest begins on 2000-01-03 and ends before the live-shadow boundary on 2026-07-24. Historical traces are visually separated from genuine future live-shadow observations. Research found positive cross-sectional Rank IC at six and twelve months, but the portfolio active-return t-statistic was only 1.15; therefore the overlay is published for live shadow and capital is not approved.
7. Value Cycle Risk Management
The Risk Managed Value Cycle is a separate research-shadow portfolio. It does not alter the HML signal, sector ranking or ETF sleeve weights. At each observed month-end, it sets total equity exposure to the lower of: (1) a 12% annualized volatility target based on the prior 63 trading sessions, and (2) a trend gate based on the dynamic Sector EW benchmark versus its 200-session moving average. Both controls have a 25% defensive floor.
The signal is observed through close t, its target becomes effective at the next trading close, and that exposure applies to the following close-to-close return. The residual allocation earns the official daily Fama-French RF return. Exposure changes incur 10 basis points in addition to the 25-basis-point cost already embedded in Value Cycle.
In the locked 2000-2026 validation, maximum drawdown improved from approximately -51.4% to -19.4%, annualized volatility declined from 17.9% to 11.1%, and Sharpe increased from 0.48 to 0.58. CAGR declined from 9.3% to 8.0%. These results are historical research evidence, not a capital approval or a guarantee that future drawdowns will be contained.
8. RRG and Momentum Maps
The Relative Rotation Graph uses standardized relative-strength level and relative momentum versus SPY. It is a phase diagnostic, not the primary score. The Sector Momentum Map uses the smoothed 0–100 score on the horizontal axis and its 21-day change on the vertical axis.
9. Macro and Market Confirmation
Macro variables include rates, curve, credit spreads, volatility, oil, the broad US dollar, financial conditions and Federal Reserve balance-sheet data. Market internals include semiconductor, Nasdaq, high-beta, equal-weight, small-cap and cyclical/defensive relative ratios. These blocks provide confirmation and context; they do not override the sector score mechanically.
10. Data Sources and Freshness
ETF prices and volumes are sourced from Tiingo. Macro data are sourced from FRED, with local/FRED stitching for selected ICE BofA option-adjusted spreads. Value Cycle uses the official daily Fama-French HML factor from the Kenneth French Data Library; IWD/IWF is not used as a substitute. The live monitor reports the effective market-data date separately from the UTC generation timestamp.
11. Limitations
- ETF composition, sector definitions and industry exposures can change through time.
- Macroeconomic series may be revised and are not fully point-in-time unless explicitly reconstructed.
- Relative-momentum signals can reverse rapidly in high-volatility regimes.
- Industry ETFs have different inception dates, liquidity profiles and concentration risks.
- Scores are research classifications, not guaranteed forward-return forecasts.
- The Value Cycle history includes a 25-basis-point cost assumption, but taxes, liquidity constraints, slippage and investor-specific suitability remain outside the model.
- The Value Cycle risk overlay relies on historical volatility and a moving-average trend filter. Both can react late to abrupt shocks or repeatedly change exposure in whipsaw markets.