A4 · Pre-Catalyst Coiling

A volatility-expansion setup: a stock compressing into a tight range right ahead of a dated catalyst, primed to break.

The short code A4 is display metadata only. The canonical identifier this archetype is keyed by everywhere — config, the screener_candidates.archetype column, the scorer emit — is the descriptive string precatalyst_coiling.

A4 is live in the universe sweep: it fires and records candidates. Read the Forward evidence above before treating it as a long — the registered forward test graded its long premise anti-predictive.

Thesis

Volatility is mean-reverting and clusters in time. A stock that has gone quiet — realized vol crushed into the bottom of its yearly range, daily bars contracting — is coiling. When that compression happens directly ahead of a known, dated catalyst (an earnings date), the spring is loaded: the catalyst is the release mechanism, and a tight range guarantees a larger relative move when it resolves. The edge is not direction — it’s the volatility expansion itself. Cheap, quiet, range-bound names into a hard date tend to move more than the market is braced for.

Direction

LONG. It is classified among the long-bias archetypes, but note the directionality caveat below — the volatility edge is non-directional even though the archetype lives on the long board.

What it looks for

The evaluator (evaluate_precatalyst_coiling); thresholds from screener_config.json § archetypes.a4:

• Realized-vol tight — realized_vol_pct_1y <= rvol_pctile_max=20.0 (vol in the bottom fifth of its trailing 252-day range).
• Contraction pattern — an inside day (inside_day=1) OR an NR7 (narrowest range of the last 7 days, nr7=1).
• Catalyst proximity — an upcoming earnings date within coil_catalyst_proximity_days=30 trading days. The proximity_factor decays linearly from 1.0 at zero days to 0.0 at 30 days.
• Options IV leg (sparse by design) — when options_legs_enabled=true, reads iv_pctile (= co_options_daily.iv_rank, 0–100). Low IV percentile (cheap options) amplifies the coil score via iv_factor = 1 − iv_pctile/100. co_options_daily is watchlist-only, so for most universe names this field is absent and the leg ABSTAINS — the score falls back to rvol_factor × proximity_factor cleanly.

How it scores

The coil metric is rvol_factor × proximity_factor, where rvol_factor = 1 − rvol_pctile/100; when an IV percentile is present it’s multiplied by iv_factor too. Tighter vol, closer catalyst, and cheaper options all push the score up. The state machine:

• none — quality fail, or realized vol above the 20th-percentile threshold.
• partial — vol is tight plus either a pattern or a catalyst (but not both), or vol tight alone.
• primed — vol tight AND a contraction pattern AND a catalyst inside the proximity window.

Pitfalls / when it fails

• It resolves both ways. This is a vol/positioning setup, NOT a directional prediction. A tight range into earnings can break violently down as easily as up. Sizing and structure must account for either resolution.
• The long premise is graded anti-predictive. On the registered sector-neutral forward test A4-long excess was reliably negative at every horizon (see Forward evidence). Treat A4 as a state read, not a long entry, until a later window turns that excess positive.
• Sparse options. For non-watchlist names the IV-cheapness amplifier is unavailable, so the score reflects only vol-tightness and catalyst proximity.
• Catalyst data quality. Proximity depends on a clean forward earnings date; a stale or missing calendar entry weakens or zeroes the proximity factor.

Related

• The Screener System (vision)
• Screener Pipeline & Lifecycle (operations)
• Mechanical reference cards: Score · State · Triggers
• Sibling archetypes: A6 · Vol Mispricing · A1 · Post-Earnings Overreaction