Signal-Driven#137

LLM Sentiment-vs-Fundamentals Divergence Map

Two AI pipelines score each market in parallel: one reads social and narrative buzz, the other digs up hard facts from trusted sources. When hype has dragged the price far away from the fundamentals estimate, you bet against the hype. The edge is isolating the boring fundamentals signal that loud crowd over-reaction tends to drown out.

What you need to run it

Social/feed ingestion + sentiment-scoring agent
Fundamentals-retrieval agent with source whitelist
Divergence threshold calibrated on resolved markets
Position + risk engine to fade overextended moves

Where this applies

Markets on Polymarket where llm sentiment-vs-fundamentals divergence map is the natural play:

Will Grand Theft Auto VI release before the end of 2027?
Will Tesla stock close above $600 at any point before the end of 2027?
Will the US officially enter a recession before the end of 2027?

Capabilities this demands

Feed ingestionModel / quantData ingestionManual research

At a glance

CategorySignal-Driven

Requirements4

CapabilitiesFeed ingestion, Model / quant, Data ingestion, Manual research

VenuePolymarket (CLOB, Polygon)

Build it

Related signal-driven strategies

This is documentation, not advice. Poly Research & Robotics publishes how these strategies work because the method should be checkable — not as a recommendation to trade them. See the full strategy database (297 strategies) or the data feeds directory.