Skip to main content

Sector research

Industry notes serve two jobs: whether to allocate and the map for stock work. One page per sector; iterate as learning evolves.

1. One-page sector template

BlockContents
Value chainUp/mid/downstream, margin pool, choke points
DriversVolume, price/power, policy, technology
Cycle stageUtilization, inventory, capex phase
StructureCR3/CR5, barriers, differentiation
Key KPIsSector-specific (see table below)
Representative namesLeaders + second tier (link to equity theses)
Main risksPolicy, substitution, cycle turn
Personal callOver / market / under / avoid + why

2. Example KPIs by sector

SectorCommon metrics
ConsumerSame-store sales, channel inventory, sell-through
IndustrialsUtilization, orders, input costs
Tech / semisUtilization, ASP, inventory weeks
FinancialsNIM, NPL ratio, provision coverage
EnergyCapex, commodity price, reserve replacement
HealthcarePricing reform impact, pipeline stage, SG&A ratio

3. Lifecycle and positioning

StageTraitsTilt
IntroductionHigh growth, high failureSmall or ETF; strict cap
GrowthPenetration rampLeaders first
MatureSlower growth, cash generativeDividends + valuation
DeclineShrinking demandUsually avoid unless deep value/cleanup

4. Chain sketch habit

flowchart LR
upstream[Upstream: inputs/equipment] --> mid[Mid: manufacturing]
mid --> downstream[Downstream: brand/channel]
downstream --> user[End demand]

For each stock, mark: chain position, bargaining power, largest dependency.

5. Comparing two sectors (same decision)

CompareSector ASector B
2-yr earnings growth (range)
Valuation percentile
Policy wind
Understanding (1–5)
Biggest unknown

Rule: understanding ≤2 → express view via index only.

  • Sector underweight ≠ ban on great companies—but demand higher margin of safety or shorter hold.
  • Sector overweight still obeys sector cap.
  • After new sector page, back-link from related equity theses.

7. Maintenance

CadenceAction
Earnings seasonRefresh leader data and cycle stage
Major policyUpdate policy block within 24–48h
AnnuallyDrop stale conclusions; keep “was wrong” records