Breakout Intelligence Terminal

What qualifies as a breakout signal

1. Price closes above a defined resistance or pattern trigger.
2. Breakout distance exceeds a minimum threshold (pattern-specific).
3. Volume confirms or does not contradict the move.
4. Liquidity passes minimum tradability checks (ADV ≥ $1M / day).
5. The signal is not immediately invalidated by a reversal back below trigger.
6. Similar patterns have acceptable historical expectancy when sample size ≥ 5.

Supported breakout types: close breakout, intraday breakout, gap breakout, base breakout, range expansion, reclaim breakout, failed- break reversal, continuation breakout, earnings breakout, news- driven breakout.

Lifecycle stages

The scanner table tags each signal with one of six progress-to-target stages (the stage badge), ordered earliest → latest:

• Near starting: pattern is just emerging; no trigger activity yet.
• Just started: pattern has traction but price is still well below trigger.
• Approaching: price within ~3.5% of the trigger on the right side.
• Breaking: trigger was just cleared and the break is fresh (≤5% past trigger).
• Extended: more than 5% past trigger — chasing risk elevated; wait for a retest. Takes precedence over "near completing".
• Near completing: price within 25% of target1 (or has overshot it) — manage out, not a fresh entry.

The individual-signal detail drawer additionally maps the raw structural state into the finer nine-stage taxonomy (forming / near trigger / triggered / confirmed / extended / retesting / reclaimed / failed / matured) for stage-specific copy and tone. Both sit on top of the engine’s five-state machine (forming → triggered → confirmed → failed → reclaimed). Canonical counts: 5 engine states · 6 table stages · 9 detail stages.

Score decomposition (composite 0..100)

• 40% Breakout Quality — pattern, volume, RS, trend, compression
• 25% Trade Readiness — distance from trigger, regime, sector alignment
• 20% Historical Edge — this symbol’s own win-rate × reliability × sample size (pooled per-family calibration “pf•” and the live detector-reliability badge are separate, wider populations shown on each row)
• 15% Regime Alignment — macro regime + sector zone

The composite is then multiplied by a risk-penalty product (each ≤ 1.0) for extended setups, earnings proximity, low liquidity, weak history, gap exhaustion, and a few others — so two signals with the same raw breakout quality can have very different composites once risk is applied.

Risk flag catalogue

• Extended — price > 5% past trigger
• Low volume — volume < 1.2× 20d avg on the break
• Weak RS — relative strength score < 50
• Earnings ≤3d — gap-risk window
• Earnings ≤7d — elevated event risk
• Low liquidity — ADV < $1M / day
• Bearish regime — macro risk-off
• Weak sector — sector in laggard cohort
• Low sample — pattern n < 5
• Weak history — pattern win-rate < 40%
• Gap exhaustion — ATR expansion > 4×
• Failed — closed back below invalidation

Decision labels

• Research-worthy — composite ≥ 70, lifecycle confirmed/ triggered/reclaimed/retesting, no chasing risk.
• Watchlist candidate — composite ≥ 50, setup forming or near trigger.
• Wait for setup — forming, score < 55. Note the trigger and revisit.
• Needs confirmation — mixed read; research prompt only.
• Weak signal · avoid chasing — extended past trigger, failed, matured, or composite < 35.

Probability grade column (A / B / C / D / —)

The grade pill is a calibrated probability — a letter is shown only when an outcome-backed cell has at least n=20 historical forward outcomes (per-pattern isotonic or a populated proof cell). A is ≥70%, B ≥55%, C ≥40%, D <40%; every displayed probability is capped at 85% (empirical breakout hit-rates are never certainties).

A grey dashed “—” means insufficient calibrated sample, not a failing setup: the detector's self-assessed confidence is deliberately not dressed up as a graded probability (that would overstate certainty at n=0). The pooled “global” calibrator degenerates toward the raw score, so it is surfaced as uncalibrated rather than a percentage. Expect the column to read mostly “—” until each pattern family accumulates ≥20 forward outcomes for the current universe; that is the honest, anti-overconfidence default rather than a bug. Grades populate as the replay history backfills.

Limitations & disclaimer

Some fields (liquidity grade, spread estimate, earnings date) are surfaced only when the upstream data source provides them. Missing fields render explicit fallback labels — never fake values. Historical expectancy with sample < 5 is shown but flagged as low-sample. This page surfaces analytical signals derived from public market data; it is not personalised investment advice. Do your own due diligence and manage risk according to your situation.