Every night, Market Prism runs eight analytical layers across 172 stocks and 569,787 scored news stories — modelling the energy behind every narrative and predicting when it will exhaust. Here's how each layer works.
Patent Pending: US 63/971,470 · 63/971,478 · 63/974,154Every night, the system ingests every story being told about 172 US-listed equities across 86 scored publications. Stories are deduplicated, timestamped, and classified by lifecycle stage — from breaking news to decaying consensus. Over 569,000 stories have been captured and scored since January 2021.
Every claim in the media narrative is stress-tested against live EDGAR filings. Management teams tell stories in their MD&A sections — and reveal reality in their financial tables. This layer measures the divergence between what the media says and what the filings actually show. High divergence is a leading indicator of earnings surprises and narrative traps.
Each individual article is forensically scored for hype density, omission gaps, and narrative drift. A story that omits material risks scores differently than one that addresses them. The system tracks how article-level signals compound across a ticker's narrative landscape — identifying when coverage quality degrades before price reacts.
Not all sources are created equal. The system maintains a Bayesian reliability model for every publication in the coverage universe, updated continuously based on historical accuracy per sector. 86 publications are scored. The system learns who to trust — and whose coverage tends to precede price reversals.
When multiple independent sources converge on the same thesis within a narrow window, it looks like confirmation but often signals fragility. This layer detects coordinated narrative clusters — when analysts, media, and social channels align too perfectly, creating a single point of failure for the prevailing story.
This is the discovery at the center of Market Prism. Temporal Energy models how long a narrative can sustain price — treating stories as physical forces with measurable energy, decay curves, and half-lives. When a story's energy depletes, price corrects. When energy is rebuilding, upside follows. Across five years of validated data, stocks drop an average of 9% when Temporal Energy collapses and gain an average of 9% when it surges. This layer is what makes the system predictive, not descriptive.
This layer maps what narratives promised before earnings against what the numbers actually delivered. When the media consensus was "blowout quarter" but EPS came in flat, that gap becomes a measurable signal. The system tracks guidance accuracy, surprise direction, and how quickly narratives adjust to new information — or fail to.
All seven preceding layers converge into a single output: one verdict per ticker, every day. The synthesis layer weighs each signal, applies regime-aware calibration (bull, bear, or choppy markets behave differently), and produces a direction, confidence level, and timing window. No ambiguity — just a clear, auditable signal.
All eight layers converge into one of three actionable verdicts — driven primarily by Temporal Energy state. Not a score, not a probability — a clear structural assessment you can visualize in Signal Lab.
Our methodology is grounded in a formal research framework. Read the complete technical paper that defines narrative physics, Temporal Energy, and the eight-layer analytical architecture.