How krishivwahi/intellihedge works
IntelliHedge is a research prototype for using point-in-time news narratives to inform option-inclusive deep hedging under jump risk and transaction costs. Its primary workflow runs through either the Streamlit demo or CLI: process timestamped headlines into regime signals, simulate market and put-price paths, then train and compare hedging policies across causal-ablation arms. The repository also contains separate data-collection and parsing utilities for several news and filings sources. The main runtime relationships are supported by the README and sampled app/CLI excerpts; the data utilities are shown as independent components because their wiring to the hedging workflow was not established.
Main entry points: news_regime.py News Regime Engine · app.py Streamlit Demo · main.py CLI Runner · merton_env.py Jump Market Simulator
Components 19
Run Interfaces
- Streamlit Demo
app.py - CLI Runner
main.py
News Perception
- News Regime Engine
news_regime.py - Headline Deduplication
news_regime.py - Narrative Clustering
news_regime.py - Regime Signal
news_regime.py
Market And Hedging
- Jump Market Simulator
merton_env.py - Put Pricing
merton_env.py - Deep Hedging Ablation
deep_hedger.py
Data Preparation
- Alpaca Collector
data_pipeline/alpaca_benzinga/alpaca_collector.py - Alpaca News Cleaning
data_pipeline/alpaca_benzinga/clean_alpaca_news.py - FNSPID Collector
data_pipeline/fnspid/fnspid_collector.py - FNSPID Parser
data_pipeline/fnspid/fnspid_parser.py - FNSPID Date Helpers
data_pipeline/fnspid/fnspid_dates.py - GDELT Collector
data_pipeline/gdelt/gdelt_collector.py - GDELT Parser
data_pipeline/gdelt/gdelt_parser.py - SEC EDGAR Collector
data_pipeline/sec_edgar/sec_edgar_collector.py - SEC EDGAR Parser
data_pipeline/sec_edgar/sec_edgar_parser.py
Other components
- Researcher
Connections 13
- Researcher → Streamlit Demo: runs demo
app.py - Researcher → CLI Runner: runs CLI
main.py - Streamlit Demo → News Regime Engine: processes headlines
app.py - Streamlit Demo → Jump Market Simulator: simulates paths
app.py - Streamlit Demo → Deep Hedging Ablation: runs ablation
app.py - CLI Runner → News Regime Engine: runs perception
main.py - CLI Runner → Jump Market Simulator: simulates paths
main.py - CLI Runner → Deep Hedging Ablation: runs ablation
main.py - News Regime Engine → Headline Deduplication: filters headlines
news_regime.py - News Regime Engine → Narrative Clustering: clusters narratives
news_regime.py - News Regime Engine → Regime Signal: emits signals
news_regime.py - Jump Market Simulator → Put Pricing: prices puts
merton_env.py - FNSPID Parser → FNSPID Date Helpers: uses date helpers
data_pipeline/fnspid/fnspid_parser.py