News Signal Dashboard
Not loaded
Live updates waiting
Token not set
Event Summaries
Luna vs Terra Model Experiment
Runs the same frozen cohort of 1,000 matured articles through Luna medium first and Terra low second. Experiment results are isolated from production predictions and validated against 12h, 24h, 48h, and 1w market movement.
No experiment has been created.
Codex Authentication
Select the current
auth.json file to restore research workers.Recent Jobs
Archived Failures
Article Impacts
Source Coverage and Prediction Movement
Ticker Pipeline Health
Article and Ticker Acquisition
Current day
Article Relationships compares signed ticker price movement with article content, structure, source and timing. Bullish and bearish labels are ignored. Results are exploratory associations and may reflect confounding, repeated coverage and multiple comparisons rather than causation.
Article and Market Relationship Explorer
Movement Over Time
Select a relationship below to compare its daily movement.
All-Horizon Combined Relationships
Numeric Correlations
Grouped Relationships
Words and Phrases
Equal $1 Ticker Culling buys USD 1 of every actionable ticker attached to every article at its stored publication-time baseline price. Each article-ticker pair remains a separate fractional-share lot. When less than USD 1 remains in cash, the lowest-value open lot is sold first; additional low-value lots are sold only when needed to fund the next purchase. A configurable holding limit also closes each lot at its first stored price on or after that age. The chart is a historical data simulation and excludes commissions, spread, slippage, FX, tax, dividends and market-hours execution constraints.
Equal $1 Ticker Culling
Changing either control reruns the same chronological history. The default holding limit always matches the age of the oldest call; shorter limits realise each lot sooner.
Every plotted point is the simulated portfolio percentage change from its initial balance at the end of a Brisbane calendar day.
Prediction Accuracy tracks every bullish or bearish ticker prediction against real market movement. Price collection continues for every call at 12h, 24h, 48h, 1w, 2w, 1m, 3m, 6m, 1y, 2y, 3y, and 4y. Every call remains an independent observation for its full lifetime; later calls do not alter or exclude earlier results.
Accuracy by Interval and Confidence
Average Call Movement Over Time
Prediction Outcomes