Corporate Actions and Their Effect on Historical Prices

Stock splits, dividends, and mergers change a historical price series in ways a backtest must adjust for, and getting the adjustment wrong is easy.

Corporate actions are events such as stock splits, cash and stock dividends, spinoffs, and mergers that change the number of shares outstanding, the reported price, or the identity of the security itself, and a historical price series that is not correctly adjusted for these events will produce false price jumps and mask real ones, which is a direct source of error in any backtest built on top of it.

A two for one stock split doubles the number of shares outstanding and halves the reported price overnight, with no change in the company's actual value. An unadjusted price series shows this halving as a 50 percent single day price drop, which a naive strategy could misread as a genuine crash, either triggering a stop loss it should not have triggered or, worse, being interpreted by a mean reversion signal as a large buying opportunity that does not actually exist. A properly split adjusted series rescales every price before the split date so the entire series is continuous in split adjusted terms, and any backtest running on raw, unadjusted prices for a universe that includes any stock…

A cash dividend reduces a stock's price by approximately the dividend amount on the ex-dividend date, since the company's cash has left the business and gone to shareholders. A price series that is not dividend adjusted will show a small, real price drop on every ex-dividend date across the full history, which for a high dividend yielding stock held over many years compounds into a meaningfully understated total return if a strategy's performance is measured only on price changes rather than total return including dividends received. Backtests intended to represent an investor's actual realized return should use a total return series, which adds dividends back into the price path, rather…

The distinction matters most for strategies holding positions for months or years rather than days, since the dividend contribution to total return compounds meaningfully only over longer holding periods, while a short holding period strategy is affected mainly through the mechanical price drop on the ex-dividend date itself rather than through the compounding effect.

A spinoff distributes shares of a newly independent subsidiary to existing shareholders of the parent company, splitting one security's history into two going forward. A backtest universe that fails to add the spun off entity as a new position, and fails to adjust the parent's price to reflect the value that left the company at the moment of the spinoff, misrepresents both the parent's subsequent price history and the total return an actual shareholder received across the combined position. Vendors vary meaningfully in how completely and how promptly they capture spinoff data, and a universe built from an incomplete corporate actions feed will simply be missing some spinoffs entirely.