Software
Collection of Reusable Code to Develop and Test Your Investment Ideas.
Data Curator
A flexible open-source Python library for retrieving, validating, and homogenizing financial data from multiple providers.
Build custom feature calculations with simple Python functions and generate clean, point-in-time datasets ready for backtesting and research. Tag homogenization and automatic data validation dramatically reduce the time required to onboard new data vendors.
Run standalone, configure via Excel, or embed into larger systems. Output to CSV, Parquet, or seamlessly to Pandas DataFrames, PyArrow tables, or custom ETL pipelines.
Spend less time integrating data. Focus on alpha.
Data Weaver
It takes curated inputs and systematically transforms them into higher-order, research-grade features by combining multiple signals across time, assets, and cross-sections. This logic enables the creation of new features by ingesting outputs from the Data Curator and enriching them with additional datasets such as historical sector classifications, macro indicators, or alternative data.
The library supports N×N cross-sectional features, rolling transformations, relative metrics, and interaction-based features, allowing researchers to build complex, reusable features in a scalable and repeatable manner.
Data Analyzer
It is designed to understand raw data and curated features before they are promoted into downstream research or strategy components. This library supports exploratory data analysis, large-scale outlier detection, and pattern discovery across extensive universes of tickers and features to evaluate stability, robustness, and economic intuition.
It produces interactive dashboards and systematic outlier reports that identify anomalous behavior at the feature, ticker, and cross-sectional level, enabling researchers to validate data quality and feature behavior without modifying the underlying datasets.
Portfolio Construction
This library combines curated features, alpha signals, risk factors, and portfolio benchmarks into investable portfolios through explicit selection, sizing, and timing decisions. It defines universe selection rules, security ranking and selection logic, position sizing methodologies, and rebalancing and execution timing (calendar-based, dynamic, or hybrid), together with portfolio constraints.
Multiple allocation frameworks are supported, including ranking-based weighting, mean–variance optimization, Black–Litterman, Hierarchical Risk Parity (HRP), and other systematic weighting and risk allocation techniques. The output consists of fully specified historical portfolios and investment universes, serving as the structural bridge between discretionary design and systematic implementation.
Backtest Engine
Designed to provide a transparent and reproducible environment for testing portfolio strategies over historical data. It supports custom rebalancing logic, dynamic weight allocation, execution price configuration (e.g., VWAP, adjusted close), and commission modeling.
The engine separates concerns between data input, order execution, portfolio valuation, and performance tracking, allowing users to plug in their own decision-making models while maintaining control over capital management and trading assumptions.
Attribution Analysis
Allows analysts and researchers to decompose portfolio returns into meaningful sources of value added. It supports return attribution across sectors, industries, factors (e.g., valuation, quality), or custom-defined groupings.
Compatible with both absolute and relative performance contexts, the tool provides granular insights into allocation effects, selection effects, and timing contributions. With flexible input formats and support for cross-sectional and time-series views, it integrates smoothly into the broader research and portfolio evaluation process.
