Pipefolio: A Pipeline-Oriented Portfolio Optimization Framework

About the project

Pipefolio is a Python-based portfolio optimization framework designed with a pipeline-oriented approach. While several excellent Python libraries for portfolio optimization already exist—such as PyPortfolioOpt, Riskfolio-Lib, and skfolio—Pipefolio takes a different direction.

Built on top of scikit-learn, Pipefolio leverages existing optimization algorithms while minimizing boilerplate code through the use of the pipe (|) operator, enabling a streamlined and efficient workflow.

The framework provides the following types of models, each serving a specific purpose in the portfolio optimization process:

• Selector
  Screens assets to narrow down the investment universe.

• Transformer
  Transforms and preprocesses data to ensure compatibility with optimization models.

• Optimizer
  Implements portfolio optimization algorithms to generate optimal allocations.

• Adaptor
  Extracts and formats the optimized weights for further analysis or implementation.

Getting started

Installation

pip install pipefolio

Documentation

Usages

This example download stock data using yfinance.

import yfinance as yf
from pipefolio.adaptor import WeightMappingAdaptor, WeightArrayAdaptor
from pipefolio.data import DataPlaceHolder
from pipefolio.enums import RiskMetric, SelectMode
from pipefolio.optimizer import (
    EqualWeightOptimizer,
    InverseVolatilityOptimizer,
    RiskAverseOptimizer,
)
from pipefolio.selector import (
    AnnualVolatilitySelector,
    AverageTradeValueSelector,
    TotalReturnSelector,
)
from pipefolio.transformer import PriceToReturnTransformer

# download data via yfinance
data = yf.download(
    ["AAPL", "AMZN", "NVDA", "BRK-B", "KO", "JNJ", "TSLA", "GS", "WMT", "MCD"],
    period="1y",
    interval="1d",
)

# data must be wrapped by DataPlaceHolder
data = DataPlaceHolder.from_dataframe(data)

# pipeline -> return optimized weights
(
    data
    | AverageTradeValueSelector(5)  # selector
    | PriceToReturnTransformer()  # transformer
    | InverseVolatilityOptimizer()  # optimizer
    | WeightMappingAdaptor()  # adaptor
)
# {'NVDA': np.float64(0.10316247719888795),
#  'TSLA': np.float64(0.08487498757904818),
#  'AAPL': np.float64(0.24167068511508258),
#  'AMZN': np.float64(0.19489272842965072),
#  'BRK-B': np.float64(0.3753991216773305)}


# chaining multiple selectors
(
    data
    | AverageTradeValueSelector(8, SelectMode.LARGEST)
    | TotalReturnSelector(5, SelectMode.LARGEST)
    | AnnualVolatilitySelector(3, SelectMode.SMALLEST)
    | PriceToReturnTransformer()
    | EqualWeightOptimizer()
    | WeightMappingAdaptor()
)
# {'WMT': np.float64(0.3333333333333333),
#  'GS': np.float64(0.3333333333333333),
#  'AMZN': np.float64(0.3333333333333333)}


# with different adaptor
(
    data
    | PriceToReturnTransformer()
    | RiskAverseOptimizer(RiskMetric.STANDARD_DEVIATION, risk_aversion=2)
    | WeightArrayAdaptor()
)
# array([6.38077192e-02, 1.94597392e-02, 1.59637665e-01, 9.15442233e-03,
#        1.94553086e-01, 2.87905182e-01, 3.25055500e-02, 6.43214713e-02,
#        2.21196278e-09, 1.68655163e-01])

Supported Models

• Selector
  • AnnualVolatilitySelector
  • AverageTradeValueSelector
  • TotalReturnSelector
• Transformer
  • PriceToReturnTransformer
• Optimizer
  • EqualWeightOptimizer
  • InverseVolatilityOptimizer
  • MaxRatioOptimizer
  • MaxReturnOptimizer
  • MinRiskOptimizer
  • RiskAverseOptimizer
• Adaptor
  • WeightArrayAdaptor
  • WeightMappingAdaptor

License

Distributed under the MIT License. See LICENSE for more information.

Maintainers

pipefolio is currently maintained by kfuangsung (kachain.f@outlook.com).

Acknowledgments

• skfolio: Python library for portfolio optimization built on top of scikit-learn.
• PyPortfolioOpt: Financial portfolio optimisation in python, including classical efficient frontier, Black-Litterman, Hierarchical Risk Parity.
• Riskfolio-Lib: Portfolio Optimization and Quantitative Strategic Asset Allocation in Python.