Academic TierA neutral framing for module hit-rate quality bands (Tier 1 through Tier 4+), used in place of marketing claims. Bands describe historical accuracy ranges relative to a 50% probabilistic baseline.Brier ScoreA measure of probabilistic accuracy based on the squared difference between an assigned probability and the final outcome. Lower is better. It evaluates probability quality, not just whether the chosen direction was correct.Brier SkillA comparison of the model’s Brier Score with a reference Brier Score, such as the market’s, on the same matched population. Positive means the model had the lower Brier Score, while negative means the reference did. The metric is separate from Directional Hit Rate.Chosen-Side ProbabilityThe probability assigned to the YES or NO side selected by a Published Signal. Its average can be compared with the realized Directional Hit Rate on the same resolved population to assess whether confidence has been too high or too low.ClusterThe set of live signals sharing the same event family (ticker prefix). Clusters help expose structurally related opportunities.Continuous LifetimeThe duration a signal has remained continuously published without dropping out of the gate stack.Daily DigestA dated public summary of QSE activity and resolved market outcomes. It provides context for that day’s record and does not replace the complete Signals or Track Record pages.Directional Hit RateThe percentage of resolved Published Signals whose published YES or NO direction matched the final market outcome. It measures directional correctness, not probability calibration, and pending or void records are excluded.Early WindowA published signal in the early portion of its time-to-close range. Typically the most time remains for the market to revise.EdgeThe uncertainty-adjusted difference between the Model Estimate and Market Estimate for a binary outcome. It is published as a percentage with a YES or NO direction and is not a guarantee or recommendation.Edge PersistenceHow long a signal has continuously survived publication thresholds. Higher persistence indicates an edge that has repeatedly cleared every gate over time.Expected Calibration Error (ECE)A summary of the count-weighted gaps between stated confidence and observed frequency across probability groups. Lower values indicate smaller calibration gaps, but the result depends on the sample and grouping used.First SeenThe earliest publication observation recorded for a given market-and-direction edge within the lookback window.Late WindowA published signal close to settlement. Less remaining time for the market to revise toward fair value.LifecycleWhere a Published Signal sits within its time-to-close progression, classified as Early Window, Mid Window, or Late Window. Late-window signals have less remaining time for the market to revise toward fair value.Liquidity ContextThe supporting context for a signal's marketability, translating raw liquidity observations into human-readable wording for the Power decision layer.Marketability (signal)A qualitative assessment of how easily a published edge can be acted upon, derived from observed liquidity and depth context. Not a price guarantee. Distinct from Marketability (Market Landscape), which is a share of observed order books.Marketability RateA normalized measure of how marketable a signal currently is, surfaced in the Opportunity Radar drawer as wording such as "Highly marketable" or "Thin liquidity".Mid WindowA published signal partway through its time-to-close range.MispricingThe raw difference between the Model Estimate and Market Estimate. Unlike Edge, it does not include QSE’s adjustment for Unknown confidence, so the two values can differ.ModuleA specialized QSE evaluation pipeline for a specific domain (e.g., NBA, NFL, MLB, NCAA MBB/WBB, Soccer, Economics, General, Chemical Exposure). Each module owns its own routing, factors, and calibration.Module Performance ContextThe historical hit-rate band for the module a signal came from, shown alongside the signal so users can judge it in module context. Power-tier surface only.Opportunity Intelligence LayerThe Power-tier decision layer on top of published signals: Opportunity Radar, edge persistence, lifecycle, marketability, module performance, related opportunities, and PEM context.Opportunity RadarLive Power-tier visualization of every currently-published Edge Spotter signal. Radius encodes time-to-close, dot size encodes edge magnitude, color encodes YES/NO direction, and the spoke encodes the source module.Position Exit Manager (PEM)The internal review loop that re-evaluates active positions and emits Hold / Trim / Exit decisions with reason codes. The Opportunity Radar surfaces the latest PEM decision context for transparency, not as a recommendation.Position ReviewA single PEM evaluation of an active position, including a decision (Hold / Trim / Exit) and structured reason codes.Probability CalibrationHow closely stated probabilities match observed outcome frequencies across many resolved cases. For example, well-calibrated 70% estimates should resolve YES about 70% of the time over a sufficiently large comparable sample.Published SignalA candidate that automatically passed every Edge Spotter publication gate (liquidity, freshness, calibration, edge cap, suppression) at evaluation time, before the market outcome was known. Publication is rules-based, not outcome-based: published signals stay in the permanent record regardless of whether they later resolve CORRECT or INCORRECT. Only published signals appear on public surfaces and in the Opportunity Radar.Related OpportunityAnother currently-published signal that shares the same event family (ticker prefix) or module as the selected signal. Surfaced in the Opportunity Radar drawer.Reliability DiagramA chart comparing average stated probabilities with observed outcome frequencies across probability groups. Values closer to the equal-frequency reference line indicate better calibration for that sample.Republication CountThe number of times a signal has reappeared in the published set after previously dropping out. Higher counts can indicate edges that the market keeps re-mispricing.Resolved SignalA Published Signal whose market has closed and settled with a final YES or NO outcome. Resolved signals are counted as correct or incorrect in directional performance; pending and void records are excluded from that calculation.Signal DirectionThe YES or NO side identified by a Published Signal. It records which side the model favored relative to the market at publication time, not a promise about the final outcome.Track RecordThe public record of eligible Published Signals and their later outcomes. It reports directional performance from the permanent publication ledger without removing losses after the fact.