How fengabing0207-oss/thesisboard works
ThesisBoard is a Streamlit research dashboard, not a trading or execution system. Researchers record pre-trade theses, inspect market headlines, or use News Signal Lab to capture headlines and assess point-in-time readiness, then audit forward outcomes and candidate strategies. Validation Lab demonstrates horizon-based and benchmark-adjusted evaluation. The graph preserves these distinct product branches and the scheduled research cycle. Some connections are based on README descriptions or the SOURCE INDEX; unsampled implementation details are not inferred.
Main entry points: app.py Streamlit pages · demo_validation_data.py Validation Lab data · research_automation.py Research cycle · news_signal_lab.py Signal feature lab
Components 24
Research dashboard
- Streamlit pages
app.py - Pre-Trade Check
src/pre_trade_check.py - Thesis journal
src/journal.py - Market News
src/market_news.py
News capture
- Headline collector
src/news_collector.py - News and audit store
src/news_store.py - PostgreSQL backend
src/postgres_news_store.py
Validation and modeling
- Signal feature lab
src/news_signal_lab.py - Point-in-time audits
src/news_signal_validation.py - Model selection policy
src/news_signal_policy.py - Position-book backtest
src/news_signal_backtest.py - Locked StrategySpec
src/strategy_spec.py - Price provider
src/price_provider.py - Validation Lab data
src/demo_validation_data.py - Event study
src/event_study.py - Abnormal returns
src/abnormal_returns.py - Forward outcomes
src/forward_tracker.py - Signal outcome store
src/signal_store.py - Rule validator
src/rule_based_validator.py - Cohort evaluator
src/signal_evaluator.py
Research operations
- Collection readiness
src/research_readiness.py - Research cycle
src/research_automation.py
Other components
- Researcher
- Anthropic summary
Connections 28
- Researcher → Streamlit pages: uses
README.md - Streamlit pages → Pre-Trade Check: renders
app.py - Pre-Trade Check → Thesis journal: saves decisions
README.md - Streamlit pages → Market News: renders
README.md - Market News → Anthropic summary: optionally summarizes
README.md - Streamlit pages → Headline collector: offers capture
README.md - Headline collector → News and audit store: records captures
scripts/collect_news.py - News and audit store → PostgreSQL backend: uses backend
src/news_store.py - Streamlit pages → Collection readiness: shows readiness
app.py - Streamlit pages → Signal feature lab: shows lab
README.md - Signal feature lab → News and audit store: reads headlines
- Signal feature lab → Point-in-time audits: audits predictions
src/news_signal_lab.py - Model selection policy → Point-in-time audits: checks leakage
src/news_signal_policy.py - Position-book backtest → Model selection policy: evaluates candidates
src/news_signal_backtest.py - Position-book backtest → Locked StrategySpec: locks strategy
src/news_signal_backtest.py - Research cycle → Headline collector: collects headlines
src/research_automation.py - Research cycle → Signal feature lab: builds dataset
src/research_automation.py - Research cycle → Position-book backtest: calibrates candidates
src/research_automation.py - Research cycle → News and audit store: records audit
src/research_automation.py - Research cycle → Collection readiness: checks progress
src/research_automation.py - Streamlit pages → Validation Lab data: shows validation
README.md - Validation Lab data → Price provider: loads prices
src/demo_validation_data.py - Validation Lab data → Event study: computes events
src/demo_validation_data.py - Validation Lab data → Rule validator: classifies signals
src/demo_validation_data.py - Validation Lab data → Cohort evaluator: compares cohorts
src/demo_validation_data.py - Validation Lab data → Forward outcomes: tracks outcomes
src/demo_validation_data.py - Forward outcomes → Abnormal returns: adjusts returns
src/forward_tracker.py - Forward outcomes → Signal outcome store: records outcomes
src/forward_tracker.py