Methodology

What a setup is

Before the open, the premarket scan reads gapping, high-volume stocks and scores each one on nine components. Core: gap size, volatility (ATR), VWAP structure and market regime. Enhancers: relative volume, options activity, short availability, news and the symbol's own history. Missing data doesn't drag a score down; it lowers the data-coverage figure shown next to it. Setups are tiered Elite or Watchlist by score and data coverage.

Each setup has three published levels: the decision level (where the idea is entered), the invalidation level (where it is wrong) and the next level (the objective). They never change after publication.

What gets published

• The morning list is every setup issued by 8:15 ET whose levels are consistent with its direction and that carries no exclusion. The free picks are the highest-scoring long and short in it.
• Setups issued after 8:15 appear in the Record (and for Pro, as intraday additions) with their own time.
• Every issued setup is in the Record, including ones logged with inconsistent levels, which are never included in a published list.

How outcomes are scored

Scorer gap_elite_60m_v2, the same rule for every setup:

• Window: starts at the first regular-session minute at or after publication — 9:30 ET for premarket setups — and lasts 60 minutes (cut at the close).
• Start price: the first bar's open, the price actually available. If it's already past the invalidation level the setup is recorded as invalidated before the open; if it's already past the next level, as moved past the next level before the open. Neither counts as a win or a loss.
• Then: invalidation level touched → exit there (or at the bar's open if it gapped through); next level touched → exit there; both in the same bar → counted as invalidation. Otherwise the exit is the last close in the window.
• R = the move from the start price to the exit, divided by the published risk (decision level to invalidation level).
• Fewer than 75% of the window's bars available → unresolved. Bars are 1-minute where available, otherwise 5-minute (noted per setup). No commissions, slippage or fill modelling.

Full specification: docs/OUTCOME_SCORING_SPEC.md. When the method changes we publish what changed; past outcomes are re-computed under a new version, never edited.

What we don't claim

No statistic on this site is a promise. Averages come from the complete Record with their sample size, and small samples mean little.