Event-Driven Strategy

Purpose

Uses event information such as news, announcements, and macro policy updates. The LLM analyzes sentiment and impact magnitude to generate event-driven trading signals. Event data is managed in CSV format, and technical signals are combined with event signals through weighted aggregation to form the final trading decision.

Workflow

1. Data collection: use the read_url tool to fetch the full text of news and announcements
2. LLM analysis: the LLM reads the news and scores it from -1.0 to 1.0 with a standardized prompt (extremely bearish to extremely bullish)
3. Generate the event CSV: write data in the date,event_type,score,source,summary schema
4. Signal aggregation: signal_engine.py reads the event CSV, applies time decay, and combines it with the technical signal

Key principle: the event CSV is the data layer, and signal_engine.py is the logic layer. Keep them decoupled.

Event CSV Schema

date,event_type,score,source,summary
2024-01-15,earnings,0.8,read_url,Q4 revenue beat expectations by 30%
2024-01-20,macro,-0.5,read_url,Central bank raised rates by 25bp
2024-02-01,policy,0.3,read_url,New-energy subsidies extended
2024-02-10,sentiment,-0.7,read_url,Bearish sentiment surged on social media
2024-03-05,insider,0.4,read_url,CEO bought 5 million shares

Field descriptions:

Field	Type	Description
date	str (YYYY-MM-DD)	Date when the event became knowable (publication date, not occurrence date. If released after market close → use the next trading day)
event_type	str	earnings / macro / policy / sentiment / insider / technical_break
score	float	-1.0 ~ 1.0 (standardized LLM score)
source	str	Data-source tag (such as read_url)
summary	str	Event summary (one sentence, no commas)

Event Type Details