Short interest

Definition

Short interest is the total of shares sold short and not yet covered — the market’s measured stock of open decline theses. It reads two opposite ways at once: as INFORMATION (shorts are documented informed traders on average — heavily shorted stocks underperform in the academic record) and as FUEL (crowded shorts are forced buyers in rallies — the squeeze). Which reading dominates is a function of crowding, borrow supply, and catalyst proximity, not preference.

How it works / structure

• The data (engine-executable): exchange/FINRA short interest is reported TWICE-MONTHLY with a lag — the official but stale series; days-to-cover (short interest ÷ average volume) is the standard crowding normalization; short % of float the other; daily vendor estimates and borrow-market data (ms-short-locate-borrow fee and utilization) are the fresher proxies, labeled by source.
• The informed-shorts evidence: Asquith-Pathak-Ritter (2005) and related literature — high short interest predicted underperformance, CONCENTRATED where institutional ownership was low (i.e. where borrow supply constrained arbitrage); shorts as a group carry information.
• The squeeze mechanics: high days-to-cover + tightening borrow + a price catalyst = forced covering into thin supply; the 2021 single-name episodes are the era’s permanent exhibit (documented in SEC’s GameStop staff report); squeezes are LIQUIDITY events, not valuation events.
• Signal split: LEVEL (crowding gauge) vs CHANGE (thesis formation/covering flows) — the platform tracks both.

When it applies

Short-side sizing discipline (entering a short that is already crowded pays worse borrow and carries squeeze tail — strategy-short-selling parameters); long-side contrarian setups (crowded shorts + improving fundamentals as a squeeze- fragility screen — with the honesty that most heavily-shorted names deserve it); event positioning where short cohorts are forced participants (event-ipo-lockups, M&A bids).

Risk profile & failure modes

• Staleness trading: the official series is up to weeks old at publication; squeeze conditions form and resolve inside the reporting gap — fresh borrow data or humility.
• Both-readings error: citing informed-shorts evidence to short a crowded name (the evidence is about the CROSS- SECTION, the squeeze is about YOUR entry) — the two frames answer different questions.
• ETF/arb contamination: short interest includes hedged and arbitrage shorts (convertible, merger, ETF create- redeem) — headline numbers overstate directional conviction on names with active arb.
• Squeeze romanticism: squeeze hunting without borrow data and catalyst dates is lottery-ticket purchasing; documented squeezes are rare relative to crowded shorts that simply grind lower.

Evidence & limits

Reporting mechanics are FINRA-documented; Asquith-Pathak-Ritter (2005) anchors the informed-shorts evidence; the SEC’s 2021 staff report documents squeeze mechanics in the modern retail-flow era. Vendor daily estimates carry unpublished error; the platform labels source and staleness on every short-interest input.

Falsifiable-thesis examples

Illustrations only, not signals:

• “The top-decile short-interest cohort (low institutional ownership) will underperform its sector over the next two quarters” — falsified by the cohort return.
• “X (days-to-cover > 8, borrow fee rising) will experience a 15%+ up-move within a quarter on any positive catalyst (fragility thesis)” — falsified at the mark.

Cross-references

• The plumbing: ms-short-locate-borrow (fees, recalls, locates)
• The strategy exposed: strategy-short-selling
• The flow context: sent-news-social (retail attention), sent-fund-flows
• Sizing the tail: risk-scenario-analysis