Guidance changes & preannouncements

Definition

Preannouncements are off-calendar guidance events — a company updating (usually cutting) its outlook BEFORE the scheduled earnings date. Skinner (1994) documented the governing asymmetry: firms preannounce bad news disproportionately (litigation-risk management — getting ahead of the miss), so the unscheduled update is itself a signal class skewed negative, arriving without the priced IV protection scheduled events carry. Mid-quarter updates, withdrawn guidance, and conference-appearance revisions form the same family.

How it works / structure

• The asymmetry mechanics (Skinner + successors): bad-news preannouncements outnumber good-news ones by multiples; the legal incentive (disclosure-timing liability) makes early warning rational for misses and unnecessary for beats — the market therefore reads SILENCE into the scheduled date as mildly positive, and an unscheduled 8-K as presumptively negative until opened.
• The unpriced-event problem (engine-relevant): scheduled earnings carry elevated IV (opt-implied-volatility event bumps); preannouncements arrive at CALM IV — the same magnitude surprise produces a larger options repricing and gap-through-stops risk without the term-structure warning (event-earnings priced-move discipline has no benchmark here).
• The severity ladder: guidance trimmed (routine) → guidance cut materially → guidance WITHDRAWN (the documented distress/uncertainty tier — withdrawal waves in March 2020 were regime information) → CFO departure adjacency (fa-earnings-quality markers).
• Reading structure: cut composition (demand vs cost vs one-time), the reaction decode (qualitative-analysis — a flat close on a guide-down is positioning information), and the cohort read (sector-wide preannouncement clusters as macro nowcasting — documented in recession onsets).

When it applies

Every single-name position (the unscheduled-event class is part of holding equities — sizing, not prediction, is the control); guidance-credibility grading per management (fa-guidance-estimates walk-down machinery extends here); sector nowcasting from preannouncement breadth; post-preannouncement drift theses (the revision cascade it triggers — sent-analyst-revisions).

Risk profile & failure modes

• No-warning gaps: the defining risk — stops and hedges calibrated to scheduled-event calendars miss the off-calendar class entirely; mgmt-stop-loss gap caveats apply at full force.
• Silence over-reading: “no preannouncement = clean quarter” is a probabilistic lean, not a rule; companies differ in preannouncement policy, and policy changes are themselves unannounced.
• Kitchen-sink ambiguity: new-CEO guide-downs bundle real deterioration with expectation-resetting (documented big-bath behavior) — the composition read separates them.
• Cluster misattribution: sector clusters can be one supplier’s problem echoing — the cohort read needs the supply-chain map, not just the count.

Evidence & limits

Skinner (1994) and the voluntary-disclosure literature carry the asymmetry; Reg FD sets the disclosure mechanics; withdrawal-wave episodes are public record. Drift after preannouncements follows the PEAD family evidence (event-earnings) directionally; magnitude is sample-bound. Preannouncement-policy differences across firms are observable but not centrally documented.

Falsifiable-thesis examples

Illustrations only, not signals:

• “X, having preannounced negatively, will cut again or miss at the scheduled print (first-cut-not-last thesis)” — falsified by an in-line-or-better print.
• “A sector’s preannouncement count doubling quarter-over- quarter precedes a negative sector earnings-revision breadth reading within a month (nowcast thesis)” — falsified by the paired series.

Cross-references

• The expectations layer: fa-guidance-estimates, sent-analyst-revisions
• The scheduled sibling: event-earnings
• The reading discipline: qualitative-analysis
• The unpriced-gap mechanics: opt-implied-volatility, mgmt-stop-loss