Post-Announcement Drift Analysis¶
Lag detection — instantiates Information Set Specification and Completeness Verification
Looks for predictable movement after public disclosure, suggesting delayed or incomplete incorporation.
Where a response test asks whether a signal reacted at the news, Post-Announcement Drift Analysis asks whether it keeps drifting after. Its defining move is to anchor to a discrete disclosure, then track the cumulative movement in the days and weeks that follow, in the same direction as the initial surprise. A signal that were fully incorporating the disclosure should stop moving once the window closes; if instead prices (or contract prices, or odds) continue to grind in the surprise's direction, that continuation is a predictable residual — a return you could have earned by simply betting on the sign of the surprise after everyone already knew it. Finding it, sizing it, and using it to rewrite the "already priced in" claim is the whole job. It is event-anchored, and it is about the leftover opportunity, not about whether the initial reaction happened.
Example¶
A prediction-market platform lists a contract on a state's election outcome. A well-publicized, methodologically strong poll drops on a Tuesday morning and the contract jumps a few cents within the hour — the immediate reaction clearly happened. The platform's analyst wants to know something the jump doesn't answer: is the new level already correct, or does the price keep creeping toward it for days? He groups thirty past polls by the sign of their surprise (poll better than the running average, or worse) and averages the contract's cumulative move over the following two weeks, anchored to each poll's release. The pattern is unmistakable in the aggregate: contracts drift another several cents in the surprise's direction over roughly a week before flattening. That drift is a predictable residual — buying the surprise's sign the morning after each poll would have paid. The analyst logs the revision: the platform's working claim that "public polling is reflected in contract prices within the hour" is downgraded to "…the initial move happens within the hour, but full incorporation takes about a week," and the market-quality dashboard is annotated accordingly.
How it works¶
- Group by surprise sign. Events are sorted by whether the disclosure beat or missed the prior expectation, because drift is defined relative to the surprise's direction.
- Track cumulative post-window movement. Using abnormal returns from a consumed baseline, it accumulates movement over the days/weeks after the reaction window — not the reaction itself.
- Test for predictable continuation. If cumulative drift in the surprise's direction is systematic across events, the residual is real: the information was under-incorporated at the announcement.
- Log the claim revision. The finding is written back as a longer, hedged latency for that information class, replacing the too-fast completeness claim.
Tuning parameters¶
- Drift horizon — how many days/weeks after the window to accumulate. Too short and slow drift is missed; too long and unrelated news contaminates the estimate.
- Surprise-sorting threshold — how large a surprise must be to enter the "positive" or "negative" bucket. Sharper cutoffs isolate cleaner drift but shrink the sample.
- Cost-and-risk haircut — how much transaction cost, slippage, and risk to net out before calling the drift exploitable rather than merely present.
- Revision trigger — how persistent the drift must be across events before the efficiency claim is formally rewritten rather than flagged as noise.
When it helps, and when it misleads¶
Its strength is catching the failure that a same-window response test is structurally blind to: an initial reaction that is real but incomplete, leaving a slow, tradeable residual. This is the classic post-earnings-announcement drift pattern[n1], and detecting it turns an overconfident "priced in instantly" claim into an honest, longer latency.
Its failure mode is mistaking compensation for a mispricing. A persistent post-event return may be a risk premium or may simply be too small to capture after costs, in which case calling it "incomplete incorporation" over-claims. Drift estimates are also fragile to the drift horizon and to contamination by later news. The guarding discipline is to net out cost and risk before declaring an opportunity, and to treat drift as evidence to investigate and log rather than automatic proof of inefficiency.
How it implements the components¶
residual_opportunity_probe— the predictable post-window drift is the residual opportunity it surfaces: the cumulative abnormal return earnable by trading the surprise's sign after the announcement.efficiency_claim_revision_log— a confirmed, cost-surviving drift rewrites the scoped claim for that information class to a longer, hedged latency.
It does NOT fix the event boundary or the reaction window, or score the in-window reaction — that is Event-Study Information-Response Test, via latency_and_update_window and response_completeness_test; and it does not run a general lagged-predictor regression across the whole series — that is Lagged-Response Regression. Its nearest twin is the lagged regression: drift is an event-anchored cumulative return continuing in the surprise's direction; the regression is a continuous time-series coefficient test on a lagged predictor, tied to no discrete event.
Related¶
- Instantiates: Information Set Specification and Completeness Verification — turns an over-fast completeness claim into a logged, longer latency.
- Consumes: Abnormal-Return / Residual Model supplies the abnormal-return baseline the drift is accumulated from.
- Sibling mechanisms: Event-Study Information-Response Test · Abnormal-Return / Residual Model · Arbitrage Opportunity Scan · Information-Source Inventory Matrix · Market-Microstructure Order-Book Probe · Lagged-Response Regression · Cross-Market Information-Leakage Check
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Post-Announcement Drift Analysis operates as an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution because it looks for predictable movement after public disclosure, suggesting delayed or incomplete incorporation.
Independent corroboration: The frozen evidence defines Post-Announcement Drift Analysis as 'Looks for predictable movement after public disclosure, suggesting delayed or incomplete incorporation', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Economics & Finance
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Studying delayed price or behavior movement after disclosure is rooted in financial event studies and market-efficiency research.
Related originating lineages:
- Statistics & Experimental Design — Statistics supplies time-series inference and tests of delayed or incomplete incorporation.
Review outcome: Independent reviewer agreement; high confidence.
Notes¶
[n1] Post-earnings-announcement drift is the long-documented tendency of stock prices to continue drifting in the direction of an earnings surprise for weeks after the release — a standing challenge to fast semi-strong efficiency, and the archetypal case this mechanism generalizes. ↩