Short selling

Definition

Short selling sells borrowed shares to profit from a price decline: borrow, sell, later buy back (cover) and return. It is the direct expression of a decline thesis in the stock itself, with a risk shape opposite to ownership — gains capped at 100%, losses theoretically unbounded — and a mechanical stack (locate, borrow fee, margin, recall) that is part of the strategy, not overhead.

How it works / structure

• Mechanics: Regulation SHO requires a locate before the sale; the position pays a borrow fee (hard-to-borrow names can cost double-digit annualized rates — ms-short-locate-borrow), posts margin (acct-margin-rules), owes any dividends to the lender, and can be recalled (forced buy-in).
• Parameters (engine-executable): entry condition, borrow-fee ceiling (a thesis that costs 30%/year to hold must clear that hurdle), size from risk budget with the asymmetry noted, stop policy (mgmt-stop-loss — mandatory in a position with unbounded loss), maximum holding time, cover triggers.
• Defined-risk alternatives: strategy-bear-put-spread and long puts express the same thesis with bounded loss and no borrow stack — the comparison belongs in every short thesis.

When it applies

Decline theses with a falsifier (deteriorating fundamentals with an event date, broken structures with a level), hedging (short index/ETF against longs — strategy-pairs-trading generalizes), and relative-value shorts. Crowded shorts are a distinct regime: high short interest (sent-short-interest) means squeeze mechanics can dominate the fundamental thesis.

Risk profile & failure modes

• Unbounded loss + adverse skew: stocks can rise without limit, and acquisition premiums land overnight; sizing must assume gaps, not paths (risk-fixed-fractional with reduced size vs long positions).
• Squeeze dynamics: rising prices force covering which raises prices — losses in crowded shorts are self-amplifying; borrow fee spikes and recalls arrive simultaneously (D’Avolio 2002 documents the borrow market’s stress behavior).
• Carry stack: borrow fees + dividends owed + margin interest make time itself the enemy — a “right eventually” short can lose money.
• Regulatory changes: short-sale restrictions can change mid-position (halts, circuit-breaker uptick rule under Reg SHO).

Evidence & limits

Mechanics are regulation (SEC Reg SHO). Academic evidence: heavily shorted stocks underperformed on average in historical samples (short sellers as informed traders is a robust literature finding), but the implementable version is bounded by borrow costs — D’Avolio (2002) showed the expensive-to-borrow names drive much of the paper return. Short-side “conviction” without an event falsifier is where unbounded-loss accidents live; the platform requires the falsifier.

Falsifiable-thesis examples

Illustrations only, not signals:

• “X will trade 20% lower within two quarters, with borrow cost under 5% annualized throughout” — falsified by price or borrow series.
• “Y will cut or suspend its dividend within 12 months” (a fundamental short falsifier) — falsified if the dividend is maintained.

Cross-references

• Mechanics: ms-short-locate-borrow, acct-margin-rules, ms-settlement (fails/close-outs)
• Positioning data: sent-short-interest
• Defined-risk alternatives: strategy-bear-put-spread, strategy-bear-call-spread
• Paired form: strategy-pairs-trading