Event returns are not automatically tradable returns

Define what the event window measures

An event study asks how prices behave around a specified information release. A trading study asks what a documented decision process could earn after receiving and acting on information. Those questions overlap but are not identical. A window beginning before an earnings announcement includes a price reaction unavailable to a strategy that waits for the announcement's contents.

Form 8-K includes a category for results of operations and financial condition, but a filing date is not automatically the earliest public earnings-release timestamp. The issuer may release information through another public channel first. Preserve the original release evidence, any filing evidence, and the data-arrival time separately. Uncertain ordering should narrow the claim rather than be filled with a convenient assumed time.

The overnight gap can dominate the result

Imagine a fictional stock closing at $100, releasing earnings after the close, opening the next session at $108 and closing that session at $109. The close-to-close event return is 9%. A hypothetical purchase at the next opening price followed by a sale at the next close earns about 0.926% before costs. The preceding 8% gap belongs to an earlier holder.

These two returns should not be added mechanically: 1.08 multiplied by 109 divided by 108 equals 1.09. The difference also does not imply that the opening price was executable for unlimited size. The NYSE auction documentation describes distinct order processes and deadlines. Any simulated auction participation must be consistent with the information cutoff and the order type actually assumed.

Construct comparable events and controls

Classify releases as before open, during session or after close using the relevant exchange calendar and timezone. A daylight-saving change or an early close can alter the classification. If only a calendar date is known, do not manufacture an intraday timestamp. Apply a conservative window or exclude ambiguous records under a rule declared before viewing their outcomes.

Specify whether the response is a raw return, a market-adjusted difference or a regression residual. A stock rising 9% when a matched benchmark rises 2% has a 7-percentage-point arithmetic difference, not necessarily a causal earnings effect. Other news, industry moves and risk exposures remain. Overlapping events and many companies reporting on the same day also make observations dependent.

Keep selection and execution evidence separate

Choose event inclusion rules before studying returns: first release versus revision, scheduled versus unscheduled announcements, and treatment of concurrent mergers or guidance. Record excluded events and reasons. For a strategy interpretation, state the processing delay, permissible order window, partial-fill treatment and exit rule. A historical best price observed after the event is not evidence that an order could have captured it.

Report event-window and post-decision returns in separate columns, with clearly named denominators and cost assumptions. This makes an informative market reaction visible without converting it into a profit claim. The example is deliberately fictional and contains no estimated earnings effect. It demonstrates why publication timing and executable timing belong in the method, not only in a footnote after performance.

Inspect the example

Sources & reproduction

Primary references support the definitions and method. Numerical examples are illustrative unless a source record is explicitly identified.

• SEC: Form 8-K, Item 2.02
• NYSE: Auctions

Continue the review