What this page is
This brief covers the industry risk premia output of our daily US equity factor model.
For each of the 60 GICS industries, the model estimates how much the market
paid to be in that industry on each trading day — after removing the
contribution of style factors such as momentum, size, value, beta, and quality.
The PDFs below show the cumulative rankings and the stock-level alpha leaders and
laggards within the top and bottom performing industries.
PDF
Industry Risk Premia — All 60 Industries
Summary · All 60 GICS Industries
Industry Risk Premia — Ranked Table
817 trading days · Data: 5 Jan 2023 – 6 May 2026 ·
Columns: Cumulative Premium, 20D Avg, 60D Avg, 10D Momentum, 20D Momentum, Trend
All 60 industries ranked from highest to lowest cumulative premium.
Green rows = top decile (D10) — consistently rewarded by the market.
Red rows = bottom decile (D1) — persistently penalised.
The 10D and 20D momentum columns show whether the premium is currently
accelerating (▲) or decelerating (▼).
⬇ Download PDF
TOP 3
Highest Cumulative Industry Premium — Stock Leaderboards
These industries have earned the strongest cumulative risk premia since Jan 2023 — the market has
consistently rewarded stocks in these sectors beyond what their style factor exposures
would predict. The leaderboards show which individual stocks within each industry captured the most alpha.
Rank 1 · Cumulative premium +0.758
Communications Equipment
18 stocks · 817 trading days
The single strongest industry tailwind in the sample. Stocks ranked by cumulative alpha —
the return not explained by their exposure to momentum, size, value, beta, or quality.
Top decile highlighted green, bottom decile red.
⬇ Download PDF
Rank 2 · Cumulative premium +0.635
Technology Hardware, Storage & Peripherals
22 stocks · 817 trading days
Second strongest industry premium. Includes hardware and storage names that benefited
from AI infrastructure buildout beyond what momentum and size exposures explain.
⬇ Download PDF
Rank 3 · Cumulative premium +0.396
Construction & Engineering
29 stocks · 817 trading days
Strongest non-tech industry premium in the sample — driven by infrastructure and
reshoring spend. Also currently the top 10D momentum industry (accelerating as of 6 May 2026).
⬇ Download PDF
BOTTOM 3
Lowest Cumulative Industry Premium — Stock Leaderboards
These industries have faced persistent headwinds beyond what style factors explain —
the market has structurally underpriced them relative to their factor exposures.
Even the best-performing stocks within these industries faced an industry-level drag.
Rank 58 of 60 · Cumulative premium −0.523
Textiles, Apparel & Luxury Goods
16 stocks · 817 trading days
Persistent industry headwind driven by consumer spending rotation away from discretionary
goods. Individual stocks ranked by alpha — the few outperformers here did so despite
the structural drag.
⬇ Download PDF
Rank 59 of 60 · Cumulative premium −0.545
Life Sciences Tools & Services
33 stocks · 817 trading days
Second worst industry premium. Biotech tools and CRO names have faced post-COVID
demand normalisation headwinds that the style factor model could not fully explain.
⬇ Download PDF
Rank 60 of 60 · Cumulative premium −0.556
Health Care Equipment & Supplies
56 stocks · 817 trading days
The single largest industry headwind in the full sample. 56 stocks — the largest
bottom-industry universe. Currently also the steepest 10D momentum decelerator
as of 6 May 2026.
⬇ Download PDF
PDF
Sector Rotation Signal — Feb 2026 to May 2026
Sector Momentum · US SPDR Sector ETFs vs S&P 500
Momentum Rotation — US Sectors
74 trading days shown · Feb 2026 – 14 May 2026 · Benchmark: S&P 500 (^GSPC) ·
Long window: 50D · Short window: 10D
Sector rotation chart for 11 US SPDR sector ETFs (XLY, XLP, XLE, XLF, XLV, XLI, XLB, XLRE, XLK, XLC, XLU)
plotted against the S&P 500. Each sector's trail shows the last 74 trading days of
relative ratio vs relative momentum.
Four quadrants: Leading (strong & improving) · Weakening (strong but fading) ·
Lagging (weak & deteriorating) · Improving (weak but recovering).
Per-sector drill-down pages show the normalised price chart vs benchmark with the tail window highlighted.
⬇ Download PDF — Feb–May 2026 (74 days)
Recent Window · Apr 1 – 14 May 2026
Momentum Rotation — What Moved in April–May
32 trading days shown · 1 Apr 2026 – 14 May 2026 · Benchmark: S&P 500 (^GSPC) ·
Long window: 50D · Short window: 10D
Shorter trail focused on the most recent 32 trading days — captures the post-tariff-announcement
regime shift and what rotated in and out from early April through today.
Useful for identifying which sectors gained or lost momentum in the current market environment.
⬇ Download PDF — Apr–May 2026 (32 days)
GICS Industry Groups · 25 Sectors · Weekly Data
Sector Rotation — GICS Industry Groups (6-Month Weekly)
26 weekly bars shown · Nov 2025 – 14 May 2026 · Benchmark: US Rebalance ·
Long window: 12wk · Short window: 2wk
Broader universe of 25 GICS industry groups using weekly data over a 6-month window.
Monthly milestone markers (◇) on each trail show the sector's position at the start of each calendar month
for visual intuition of how rotation evolved over time.
Includes a compact price-vs-benchmark panel per sector.
Four quadrants: Leading · Weakening · Lagging · Improving.
⬇ Download PDF — GICS 25 Sectors · 6-Month Weekly
PDF
Sector Predictions — ML Weekly Signal
GICS 25 Sectors · Random Forest · 4-Week Horizon · Weekly Data
Current Predictions — All GICS Industry Groups
25 sectors · Data through 1 May 2026 · Benchmark: US Rebalance ·
Trained on 20-year weekly history · 4-week forward horizon
Machine learning predictions (Random Forest) for each of the 25 GICS industry groups vs the US Rebalance benchmark.
Each sector is assigned a current signal (UP ↑ or DOWN ↓) with a confidence score.
Includes recent accuracy: how often each sector's model was correct over the last 52 weekly bars
(out-of-sample, no look-ahead). Signal strength chart and accuracy bar chart included.
⬇ Download PDF — Sector Predictions · 14 May 2026
METHOD
How These Numbers Were Produced
The model
Every trading day we run a cross-sectional OLS regression across
all ~1,200 US stocks in the universe:
stock_returni,t = αt + Σ βk,t · exposurei,k
+ Σ γj,t · 𝟙[industryi = j] + εi,t
- The βk,t coefficients are the daily style factor risk premia —
what the market paid per unit of momentum, size, value, beta, quality, etc.
- The γj,t coefficients are the industry risk premia —
the return advantage (or disadvantage) of being in industry j on day t,
after stripping out all style factor effects.
- The residual εi,t is the stock's alpha —
the part of its return unexplained by either style factors or industry membership.
Style factors controlled for
| Factor | What it captures |
|---|
| GEMLT_MOMENTUM | Price momentum (trailing 12-month return) |
| GEMLT_SIZE | Market capitalisation (log scale) |
| GEMLT_VALUE | Book-to-market ratio |
| GEMLT_BETA_D | Systematic market sensitivity (beta) |
| GEMLT_EARNINGS_YIELD | Earnings yield (E/P) |
| GEMLT_DIVIDEND_YIELD | Dividend yield |
| GEMLT_LEVERAGE | Financial leverage |
| GEMLT_GROWTH | Earnings growth |
| GEMLT_QUALITY | Profitability and earnings quality |
| +6 more GEMLT factors | Volatility, liquidity, mid-cap, etc. |
What "cumulative premium" means
The cumulative premium is the running sum of daily γj,t
coefficients from 5 January 2023 to 6 May 2026 (817 trading days).
A cumulative premium of +0.758 (Communications Equipment) means that — on a
hypothetical equally-weighted position in that industry — the industry dummy alone
contributed roughly 75.8 percentage points of gross return over the sample period,
net of style factor effects. This is the pure industry tailwind.
What "stock alpha" means in the leaderboard PDFs
Each stock's alpha on day t is its regression residual:
alphai,t = actual_returni,t − predicted_returni,t
The predicted return uses the style factor exposures and the cross-sectional
factor premia — it does not include the industry dummy. So the alpha absorbs
both the industry premium and the stock's idiosyncratic return.
Stocks ranked highest in the leaderboard outperformed both their factor model prediction
and their industry peers.
Data note: Factor exposures are sourced from client-supplied Barra GEMLT
model snapshots (daily). Returns are total returns (DATA_TOTAL_RETURN_B).
Same-day alignment is used (exposures and returns measured at the same close);
for production use, validate whether next-day alignment is more appropriate
to avoid any lookahead in the exposure data.