Insights

What Is Delisting Bias?

Alphanume Team · April 24, 2026

Handling terminal returns so shorts aren't overstated — and longs aren't either.

Delisting bias is the systematic distortion that arises when securities that are delisted during a backtest are handled improperly — most commonly by simply dropping out of the dataset at delisting, with no terminal return assigned. The handling matters: how the final value is set determines whether shorts and longs are evaluated fairly relative to a benchmark that includes those terminal events.

The distinction from survivorship bias

Delisting bias and survivorship bias are related but distinct:

Survivorship bias: Delisted names are excluded from the universe entirely. They were never in the sample.
Delisting bias: Delisted names are included, but their terminal returns are mishandled — typically by stopping price tracking at delisting without recording the actual final return.

A backtest can avoid survivorship bias while still suffering delisting bias if the included delisted names have incorrect terminal values.

The common error

The default behavior in many data systems: stop tracking the security at delisting, with the last price as the terminal value. Consequences:

A security that traded at $4 before suspending and then delisted to zero shows a terminal value of $4 in the data — a -50% loss looks better than the realized -100%.
A security that traded at $0.50 before delisting through bankruptcy shows -95% but the true realized loss to a long holder may include further OTC trading or zero proceeds.
A security acquired at $25 cash shows $25 as the final price (correct), but if the data stops at $20 the day before the acquisition closes, the gain to the holder is understated.

How CRSP and other vendors handle it

CRSP (the academic standard) computes delisting returns explicitly. The methodology:

For securities that continue OTC after exchange delisting, the OTC price is used.
For securities that delist through merger, the deal value is used.
For securities that delist through bankruptcy without OTC continuation, a delisting return is assigned based on subsequent recovery (often zero, sometimes a small percentage of last trading price).

The CRSP delisting return field is the canonical academic standard. Most academic event studies use it.

Why the direction of bias matters

Delisting bias generally works against short positions and in favor of long positions when handled naively:

For shorts: the final gain on a name that goes to zero is understated if the data shows the last traded price (e.g., $1) as the terminal value rather than zero. Short alpha is reduced.
For longs: the final loss on a name that goes to zero is understated similarly. Long return is overstated.

For dilution-event short research, where many of the worst-performing names eventually delist, the bias is material. Studies that handle delisting returns correctly typically report larger short-side alpha than studies that don't.

What proper handling looks like

To handle delisting properly:

Identify delisting events for each security in the historical record.
For each delisting, determine the terminal value:
- Cash acquisition: deal price.
- Stock acquisition: value of received shares.
- OTC continuation: subsequent OTC price.
- Bankruptcy without continuation: zero or recovery value.
- Voluntary delisting: subsequent post-listing price (may be near last traded).
Compute the terminal return from last traded price to terminal value.
Include this terminal return as the final observation in the security's return history.

The "delisting return" data field

Vendors with proper delisting handling provide a "delisting return" field:

For securities that did not delist in the period: null.
For securities that delisted: the cumulative return from last traded price to terminal value.

Backtests should add this delisting return to the security's last-day return in the backtest period.

Practical impact

For a representative dilution-event short strategy over a 10-year backtest period:

Approximately 10–20% of event-flagged names delist within 24 months of the event in some subsegments.
The average delisting return for these names is in the -50% to -90% range relative to the pre-delisting price.
Properly accounting for these terminal returns adds 1–4% to the average per-event return of the strategy.

Sources of delisting return data

CRSP (academic / institutional).
Bloomberg with appropriate corporate-action handling.
Compustat with linkage to delisting data.
Third-party data providers offering point-in-time-aware historical price data.

Most free / low-cost data sources do not provide delisting return data and require manual reconstruction.