heurist-finance: Skill for Claude Code

Sector Rotation

Shared plumbing and output contract: ../_shared/analyst-core.md. Sector selection explains a large share of a stock pick's return — a right stock in a wrong sector fights the tide the whole hold. Two engines drive this skill: relative strength (empirically the stronger one — momentum works notably well at the sector/industry level, where trends persist longer than in single names) and the cycle template (useful map, unreliable clock).

The relative-strength engine (primary)

Data: prices on the SPDR set (XLK XLF XLE XLV XLI XLP XLY XLU XLB XLC XLRE) vs SPY, 6mo–2y; technical_snapshot symbols=... batched; fund_snapshot for what's actually inside a sector ETF (XLK's concentration means "tech sector" ≈ three stocks — always check before treating an ETF as a diversified sector view).

• Rank sectors by 3- and 6-month return vs SPY (computed, not recalled). Industry/sector momentum was Moskowitz & Grinblatt's finding before stock momentum got the fame — persistent enough that the top-3-sectors-by-RS rule has real backing. New leadership (a sector crossing into the top ranks after quarters at the bottom) is more information-rich than incumbency.
• Confirm with breadth inside the sector: a sector rally led by one mega-cap while the median member lags is a stock story wearing a sector costume (check equal-weight variants — RSP-style — via prices, or eyeball top-holdings dispersion from technical_snapshot on the ETF's top names). Broad participation = allocatable theme; narrow = pick the stock instead, or skip.
• Duration of a rotation: sector leadership regimes historically persist quarters, not days — which is exactly the user's horizon. A two-week bounce in last year's loser sector is more often short-covering than rotation; demand a month-plus of RS improvement plus a macro reason before calling it a turn.

The cycle template (secondary — map, not clock)

The classic sequence — early cycle: financials, consumer discretionary, industrials; mid: tech, materials; late: energy, staples, healthcare; recession: staples, utilities, duration proxies. It's a real tendency with an honest caveat the user must hear: cycle phase is only identifiable with confidence in hindsight, and the sequence has broken in structurally driven markets (2020's tech-led "recession", AI capex overriding rate sensitivity). Use the template to explain and cross-check an RS reading ("energy leading + curve steepening = late-cycle-consistent"), never to front-run a phase the data hasn't confirmed.