news-event-strategy

What problem does it solve?

It converts unstructured news, announcements, and macro events into a structured, time-decayed event signal that can be combined with technical signals for trading decisions.

Core Features & Use Cases

• Event-driven sentiment scoring: Fetches full news/announcement text (via read_url) and scores impact on a standardized scale from -1.0 to 1.0.
• CSV event data layer: Produces an event CSV using the date,event_type,score,source,summary schema to keep data separate from trading logic.
• Time-decay aggregation + weighted blending: Applies exponential time decay to event scores and aggregates them into an event signal, then blends it with the technical signal using an alpha weight.
• Use case: For a backtest, maintain an instrument-specific historical event CSV and generate daily event signals that naturally remain zero on event-sparse days.

Quick Start

Create a historical event CSV in the required schema and then use signal_engine.py to compute the time-decayed event signal and combine it with your technical signal using alpha=0.6.