Bankruptcy as a Short Signal: What the Data Shows

Pre-default drift, borrow costs, and the squeeze risk.

The idea behind bankruptcy short selling is intuitive: equities of companies drifting toward insolvency should underperform as dilution, delisting, and ultimate wipeout risk accumulate in the price. The empirical record does support a pre-default drift in distressed names. But the gap between gross signal and net realized return is unusually wide here, and closing that gap honestly requires confronting a set of structural frictions that most surface-level discussions skip. For researchers approaching this as an event-driven edge, the right starting point is the corporate default events dataset, which captures filings, covenant violations, and distress indicators on a point-in-time basis.

The thesis: why bankruptcy short selling has a mechanical basis

When a company enters financial distress, several forces bear down on the equity simultaneously. Management facing a liquidity crunch will often issue dilutive equity — sometimes at severe discounts — to extend runway. Creditors begin negotiating, and in most Chapter 11 reorganizations existing equity is either cancelled or receives a negligible recovery relative to par. Credit rating downgrades trigger forced selling from mandated investors. Index exclusions remove a category of price-insensitive buyers. Each of these effects is directionally negative for the equity, and they tend to cluster in the months preceding a default event.

The pre-bankruptcy window — roughly six to eighteen months before the filing date — has received meaningful attention in the academic literature. The mechanism is not simply that the stock falls; it is that the fall occurs in a recognizable pattern tied to identifiable catalysts: missed interest payments, going-concern audit opinions, failed refinancing attempts, and distressed debt exchanges that signal creditor impatience. These are observable, structured events, which is what makes the thesis tractable for systematic research rather than purely discretionary.

The frictions that compress the edge

The gross return profile of pre-bankruptcy shorts looks compelling before costs. After costs, the picture is substantially different. Three frictions dominate:

• Borrow cost and availability. Distressed names are among the most expensive securities to borrow. Hard-to-borrow fees can run from a few hundred basis points annually to rates that exceed 100% annualized for the most distressed small-caps. These costs accrue daily while the position is open, which can span quarters. A position that earns 40% gross over six months can be cut to near zero or worse after borrow fees.
• Buy-in and recall risk. Short sellers do not own the borrow — they rent it. Prime brokers can recall shares at any time, forcing a cover at whatever price prevails. In distress situations, brokers managing risk often reduce exposure to the same names at the same time, creating simultaneous recall pressure exactly when covering is most expensive.
• Short squeeze dynamics in low-float names. Bankrupt and near-bankrupt equities are particularly prone to violent short squeezes. Float is often reduced by prior equity conversions, insider lockups, or thin post-filing trading. When a catalyst — a reorganization plan announcement, a speculative press release, coordinated retail buying — lifts the price sharply, short interest as a percentage of float means small amounts of buying pressure translate to outsized price moves. The negative expected value of these squeezes is a material drag even when the eventual outcome is correct.

The implication is that the observable edge in distressed equities is partly, and perhaps largely, a compensation for bearing these costs and risks rather than a pure alpha anomaly. A strategy that earns its Sharpe ratio by accepting illiquidity, recall exposure, and squeeze tail risk is a legitimate strategy — but it should be sized and evaluated accordingly, not treated as a free lunch.

Designing the study correctly

Survivorship and delisting bias are severe in this area. Any backtest that relies on a price database containing only currently traded securities will systematically exclude the names that went to zero — which are exactly the names the strategy is designed to capture. A correct research design requires:

1. Point-in-time universe construction. The universe at each date should reflect what was observable then, not what survived. This means including names that subsequently delisted, were acquired out of distress, or ceased trading.
2. Delisting return treatment. When a security is delisted, most price databases stop recording returns. The actual return from the last traded price to zero (or a small recovery value) must be imputed. Ignoring this step overstates the performance of any long strategy and understates the performance of any short strategy in distressed names.
3. Borrow-cost-adjusted returns. Gross short returns should be reduced by realistic borrow costs specific to each name's hard-to-borrow status at the relevant dates. Using a flat borrow-cost assumption understates the friction for the most distressed names, which are the same names with the highest theoretical gross returns.
4. Event-time alignment. Returns should be measured relative to identifiable event dates — going-concern opinion filings, credit rating actions, Chapter 11 petition dates — rather than calendar time, to avoid noise that dilutes the signal.

For a broader discussion of how to structure this type of universe, the post on how to find stocks to short sell using data covers the practical data pipeline in detail. For the underlying data infrastructure, the considerations around market data sources for systematic short-selling research apply directly here.

Position sizing for negative-skew payoffs

The payoff distribution for a pre-bankruptcy short is negatively skewed from the short seller's perspective — the frequent small gains from gradual equity erosion are interrupted by occasional large losses from short squeezes, even when the eventual fundamental outcome is a zero recovery. This is the mirror image of the lottery-ticket profile: instead of many small losses and rare large gains, you have many small gains and rare large losses.

Position sizing should reflect this. Full Kelly allocation to a strategy with this payoff shape produces ruin risk that is unacceptable in practice. A fractional Kelly approach, combined with hard position limits and pre-defined stop-loss levels that limit squeeze exposure, is more appropriate. A stop-loss that triggers at a defined percentage above entry — irrespective of the fundamental view — forces the discipline of re-entering at a higher cost rather than riding a squeeze to a margin call. The exit is a cost of doing business, not a defeat.

Structural role: a sleeve, not a standalone strategy

Pre-bankruptcy shorts work best as one component of a broader event-driven short book, not as a standalone strategy. The reasons are structural. Borrow availability constrains the investable universe to names where a short position can actually be established and maintained. Squeeze events create mark-to-market drawdowns that are psychologically and operationally difficult to hold through without a diversified book providing offsetting PnL. Transaction costs — borrow fees, spread, and recall-driven forced covers — mean the strategy needs meaningful gross returns to net anything after friction.

Within a diversified short book, the distressed sleeve adds a fundamentally different signal source from momentum shorts or dilution-event shorts. The holding period is longer, the catalysts are more credit-driven, and the risk profile is distinct. That diversification is a genuine benefit. But the sleeve should be sized to reflect its negative-skew characteristics and its friction load — typically a smaller allocation than the gross return profile would naively suggest.

Where Alphanume fits

Alphanume's corporate default events dataset provides structured, point-in-time coverage of bankruptcy filings, going-concern opinions, credit rating actions, and distressed exchanges — the event types most directly relevant to building and maintaining a distressed short universe. Events are timestamped to the filing or announcement date, not the resolution date, which is the correct alignment for trading research.

Explore the Corporate Default Events dataset →