Delisting Returns

A stock that goes bankrupt doesn't just quietly disappear from a price database — the shares typically became worthless, and a shareholder who held them lost essentially their entire position. But many price databases, especially older or lower-quality ones, simply stop showing a row for that ticker once it stops trading, with no final return recorded at all. A backtest that treats "no more data" as "the position was closed with no gain or loss" is quietly deleting the worst outcomes in the strategy's history — precisely the outcomes a risk-aware backtest most needs to capture.

The idea

A delisting return is the return that should be applied on the date a security stops trading, reflecting what actually happened to a holder: a merger or acquisition typically pays out a cash or stock amount close to the pre-announcement price (a mild, sometimes even positive, return); a bankruptcy typically wipes out equity holders almost entirely (a return close to −100%); an exchange delisting for failing listing standards often precedes a steep decline. Academic research on major US databases has documented that stocks delisted for negative reasons (bankruptcy, being dropped for poor performance) average notably negative returns in the delisting event itself, and a database that records no return at all for these — treating them as simply vanishing — introduces a measurable upward bias into any backtest built on it.

Worked example

A small-cap value strategy holds a stock that files for bankruptcy and is delisted three months later. If the backtest's data provider simply stops returning prices for that ticker, a naive backtest engine might treat the position as unchanged from its last recorded price, or silently drop it from the portfolio with no loss recorded — either way, erasing what was actually close to a total loss. A correctly constructed backtest instead applies an explicit delisting return (either a data vendor's recorded delisting price, or a conservative assumption like −100% for a bankruptcy known to have wiped out equity) on the delisting date, so the strategy's simulated portfolio actually absorbs the loss its historical counterpart would have.

What this means in practice

This matters most for strategies with meaningful exposure to distressed or small-cap names, where the rate of bankruptcies and forced delistings is far higher than in a large-cap universe — ignoring delisting returns there can turn a genuinely losing strategy into a backtest that looks profitable purely because its worst holdings vanished from the sample without penalty. Any point-in-time universe construction needs a data source that records delisting events and their associated returns explicitly, not just "last price available."