How zhaorongdai/quantlab works
quantlab is a quantitative-equity research backend: users acquire market data, build timestamp-by-symbol panels, derive factors and forward-return labels, train predictive models, and backtest resulting portfolios. Its common panel/storage layer supports that workflow, while factor analysis and run reporting provide research outputs. The README documents the stage sequence and major capabilities; source excerpts and the index support the internal storage, factor, configuration, and backtest links shown. Some implementation details were not sampled, so uncertain connections are omitted.
Main entry points: factor.py Factor layer · backtest.py Backtesting · data.py Panel datasets · model.py Models
Components 26
Data ingestion
- Acquisition
quantlab/base/acquisition.py - Vendor adapters
quantlab/acquisition
Research data
- Panel datasets
quantlab/base/data.py - Stock dataset
quantlab/dataset/stock.py - CRSP universes
quantlab/dataset/crsp - Factor layer
quantlab/base/factor.py - Factor engines
quantlab/factor - Forward labels
quantlab/label/forward.py - Factor analysis
quantlab/analysis/factor_report.py
Modeling
- Models
quantlab/base/model.py - Model implementations
quantlab/model
Portfolio evaluation
- Portfolio construction
quantlab/base/portfolio.py - Backtesting
quantlab/base/backtest.py - Vectorbt engine
quantlab/backtest/engine_vectorbt.py - Run reports
quantlab/utils/backtest_report.py
Shared infrastructure
- Panel storage
quantlab/backend.py - Pipeline configuration
quantlab/base/config.py - Data vocabulary
quantlab/enums/data.py - Experiment tracking
quantlab/base/tracking.py - Panel utilities
quantlab/utils/symbol_axis.py - Date ranges
quantlab/utils/date_range.py - Atomic JSON writes
quantlab/utils/atomic.py - JSON serialization
quantlab/utils/jsonable.py - Timing
quantlab/utils/timer.py
Other components
- Researcher
- Market vendors
Connections 36
- Researcher → Acquisition: requests data
README.md - Market vendors → Vendor adapters: provides data
README.md - Vendor adapters → Acquisition: implements adapters
quantlab/acquisition/alpaca.py - Acquisition → Panel datasets: supplies raw files
README.md - Panel datasets → Stock dataset: specializes
quantlab/dataset/stock.py - Panel datasets → Panel storage: persists panels
quantlab/base/data.py - Stock dataset → CRSP universes: uses historical membership
quantlab/dataset/crsp/__init__.py - Researcher → Panel datasets: requests panels
README.md - Panel datasets → Factor layer: provides panels
README.md - Factor layer → Factor engines: uses factor backend
quantlab/factor/kunquant.py - Factor layer → Factor analysis: analyzes factors
quantlab/base/factor.py - Researcher → Factor layer: requests features
README.md - Factor layer → Forward labels: wraps as targets
quantlab/base/factor.py - Factor layer → Models: provides features
README.md - Forward labels → Models: provides targets
README.md - Models → Model implementations: dispatches training
quantlab/model/predefined/xgb.py - Researcher → Models: trains predictors
README.md - Models → Backtesting: provides predictions
README.md - Stock dataset → Backtesting: provides prices
README.md - Backtesting → Portfolio construction: constructs weights
quantlab/base/backtest.py - Backtesting → Vectorbt engine: simulates weights
quantlab/base/backtest.py - Backtesting → Run reports: writes run report
README.md - Backtesting → Experiment tracking: logs run records
quantlab/base/backtest.py - Researcher → Backtesting: runs evaluation
README.md - Pipeline configuration → Panel datasets: configures datasets
quantlab/base/config.py - Pipeline configuration → Factor layer: configures factors
quantlab/base/config.py - Pipeline configuration → Models: configures models
quantlab/base/config.py - Pipeline configuration → Backtesting: configures backtests
quantlab/base/config.py - Data vocabulary → Pipeline configuration: defines data types
quantlab/base/config.py - Panel datasets → Panel utilities: aligns symbol axes
quantlab/base/data.py - Factor layer → Date ranges: uses date ranges
quantlab/base/factor.py - Models → Experiment tracking: records experiments
quantlab/base/model.py - Factor layer → Atomic JSON writes: persists metadata
quantlab/base/factor.py - Backtesting → Atomic JSON writes: persists run state
quantlab/base/backtest.py - Experiment tracking → JSON serialization: serializes metrics
quantlab/base/tracking.py - Factor engines → Timing: times computation
quantlab/factor/kunquant.py