Long Short Construction

Methodology fused from institutional portfolio-construction and buy-side long/short frameworks; all text is an original paraphrase.

Defaults

Preflight

Run canonical pre-flight per contracts/preflight.md. Include the X-Agentii-Trace header on every tool call per contracts/x-agentii-trace-header.md — carry the _run_id from your first tool result and name yourself (and your parent, if you were spawned).

Data Source Priority

1. Construction methodology — references/construction-methodology.md (bundled exposure framework)
2. Strategy frameworks — search_investment_strategies(domain=fundamental, kind=position_sizing)
3. Historical analogues — search_by_analogue(market_regime=...) for regime-specific exposure precedent
4. Market data — ~~market_data placeholder for beta estimation and borrow/short-interest inputs

Methodology

Retrieval Scope

structured_only

Retrieval Strategy

Ownership & insider signals: search_institutional_holdings (top-10 holders + whale portfolios, direction=accumulating|reducing|new|exited) and search_insider_trades (Form-4 transactions with SEC URLs) are available as signal inputs.

Branch (a) Structured Data Query from contracts/retrieval.md. Retrieve construction frameworks via search_investment_strategies; retrieve regime precedent via search_by_analogue. Detailed methodology in references/construction-methodology.md.

Temporal Scope

See frontmatter temporal_scope block.

Tool Allowlist

See frontmatter allowed_tools.

Protocol

Gross and net exposure are two independent dials. Gross sets how much stock-specific opportunity the book harvests; net sets how much of the return is simply the market. Raising gross while holding net constant is the defining move of long/short construction — it is what separates the strategy from levered long-only. Full derivations, attribution worked examples, and book-level limit tables are in references/construction-methodology.md.

Foundational principle: return tracks net exposure, not gross. Adding equal-beta shorts to a long book halves the return without improving selection. A short book that exists only to damp beta is a pure drag — shorts must earn their own alpha, or index-level hedging is the cheaper and more honest instrument.

Steps

1. Exposure Inventory: Compute long %, short %, gross (L+S), and raw net (L−S) against NAV. Record the starting point before any proposed change.

2. Beta-Adjusted Net (the decision-grade measure): Compute (Long% x weighted long beta) - (Short% x weighted short beta). Raw net silently assumes both sides share market sensitivity. High-beta growth longs hedged with defensive low-beta shorts can carry more directional risk than raw net implies. If raw and beta-adjusted net diverge by more than beta_net_deviation_max, the hedge is mis-specified — re-select or re-size the short side rather than reporting raw net.

3. Risk Decomposition: Confirm the book's residual is idiosyncratic. Market components offset across paired exposure, leaving long-side plus short-side company/industry risk. Note the failure mode explicitly: with beta removed there is no tailwind to carry weak selection.

4. Short-Side Classification: Separate alpha shorts (held to earn a return on their own thesis) from index shorts (held to damp beta). These are not interchangeable — using alpha shorts as a beta hedge pays the analytical cost of the former for the return profile of the latter. Screen every alpha short against the seven structural constraints (market long bias, unbounded loss, timing, borrow availability, short interest and days-to-cover, volatility asymmetry, sizing).

5. Position Drift Check: Short weights move adversely by construction — a losing short grows into the book while a losing long shrinks out of it. Schedule re-sizing rather than relying on stops alone. Flag any short exceeding max_short_position or max_days_to_cover.

6. Pair Integrity (when expressing an explicit pair): same primary risk factor on both legs; beta-match rather than dollar-match; each leg must clear the research bar independently; name the divergence catalyst and its date range; size for the decoupled case, since correlated legs decouple precisely under the stress the pair was built to survive.

7. Sensitivity Grid: Publish fund return across a −20% to +20% market range. The slope of the row is net exposure; the intercept is alpha. Diagnose both separately — a book can post a good return while its intercept is zero and its slope is merely large.

8. Limit Reconciliation: Check gross, net band, sector net, and per-position sizes against Defaults. Recompute all three exposure measures after every position change; the two dials stay independent only if measured continuously, otherwise a series of individually reasonable trades silently converts a hedged book into a levered directional one.

9. Output: Report both exposure measures, the attribution split (beta contribution vs alpha contribution per side), the sensitivity grid, and every limit breach.