PriceFrame

A modern, efficient price data management library for Python, built on top of PyArrow, Polars, and pandas.

Overview

PriceFrame provides a canonical, high-performance way to store, manipulate, and analyze financial price data (OHLCV). It combines the speed of PyArrow and Polars with the familiarity of pandas, offering a flexible feature system for technical indicators and quantitative analysis.

Key Features

• High Performance: Built on PyArrow for efficient columnar storage and operations
• Multiple Engine Support: Seamlessly switch between pandas and Polars
• Rich Feature System: Declarative technical indicators (MA, EMA, RSI, ATR, MACD, Bollinger Bands, etc.)
• Multiple Data Sources: Bloomberg, custom databases, and more
• Flexible Time Handling: Automatic timezone normalization and resampling
• Portfolio Analysis: Built-in naive portfolio construction and rebalancing
• Type-Safe: Canonical schema with automatic type casting and validation

Installation

Basic Installation

pip install priceframe

With Optional Dependencies

For Bloomberg Terminal support:

pip install priceframe[bloomberg]

For cryptocurrency data (via CCXT):

pip install priceframe[crypto]

All optional dependencies:

pip install priceframe[bloomberg,crypto]

Quick Start

Creating a PriceFrame

import pandas as pd
from priceframe import PriceFrame

# From pandas DataFrame
df = pd.DataFrame({
    'symbol': ['AAPL', 'AAPL', 'AAPL'],
    'ts': pd.date_range('2024-01-01', periods=3, tz='UTC'),
    'open': [150.0, 151.0, 152.0],
    'high': [151.0, 152.0, 153.0],
    'low': [149.0, 150.0, 151.0],
    'close': [150.5, 151.5, 152.5],
    'volume': [1000000, 1100000, 1050000]
})

pf = PriceFrame.from_pandas(df, interval='1d')

Loading from Bloomberg

# Single ticker
pf = PriceFrame.from_bloomberg(
    tickers='AAPL US Equity',
    start_date='2024-01-01',
    end_date='2024-12-31',
    interval='1d'
)

# Multiple tickers
pf = PriceFrame.from_bloomberg(
    tickers=['AAPL US Equity', 'MSFT US Equity', 'GOOGL US Equity'],
    start_date='2024-01-01',
    type_='ohlcv',
    currency='USD'
)

Adding Technical Indicators

from priceframe.features import FeatureSpec

# Define features declaratively
features = [
    # 20-day moving average
    FeatureSpec(
        name='f_ma_20',
        func='ma',
        input_col='close',
        params={'window': 20}
    ),
    
    # RSI(14)
    FeatureSpec(
        name='f_rsi_14',
        func='rsi',
        input_cols={'close': 'close'},
        params={'window': 14, 'method': 'wilder'}
    ),
    
    # ATR(14)
    FeatureSpec(
        name='f_atr_14',
        func='atr',
        input_cols={'high': 'high', 'low': 'low', 'close': 'close'},
        params={'window': 14, 'method': 'sma'}
    ),
    
    # Daily returns
    FeatureSpec(
        name='f_ret_1',
        func='ret',
        input_col='close',
        params={'horizon': 1}
    )
]

# Apply features
pf_with_features = pf.with_feature_specs(features)

# Convert to pandas for analysis
df_result = pf_with_features.to_pandas()

Data Access

# Get OHLCV for a specific symbol
ohlcv = pf.ohlcv('AAPL US Equity', engine='pandas')

# Get close prices only
closes = pf.close_series('AAPL US Equity')

# Create a close price matrix (timestamps × symbols)
close_matrix = pf.close_matrix(
    symbols=['AAPL US Equity', 'MSFT US Equity'],
    how='inner'
)

Time Operations

# Filter by date range
pf_filtered = pf.range_date(
    start='2024-03-01',
    end='2024-06-30',
    closed='both'
)

# Resample to weekly
pf_weekly = pf.resample('1w')

# Rebase to 100
pf_rebased = pf.rebase(base=100.0, anchor='close')

Portfolio Construction

# Create a naive portfolio
portfolio = pf.naive_portfolio(
    symbols=['AAPL US Equity', 'MSFT US Equity', 'GOOGL US Equity'],
    weights=[0.4, 0.3, 0.3],
    base=100.0,
    name='Tech Portfolio',
    as_='series'
)

Merging PriceFrames

# Combine two PriceFrames
pf_combined = pf1.add(pf2, require_same_interval=True)

Available Technical Indicators

PriceFrame includes a comprehensive feature registry with the following indicators:

Returns

• ret: Simple returns
• logret: Log returns

Moving Averages

• ma: Simple Moving Average (SMA)
• ema: Exponential Moving Average (EMA)

Volatility

• rolling_vol: Rolling standard deviation

Cross-Sectional

• cs_rank: Cross-sectional rank
• cs_zscore: Cross-sectional z-score

Technical Indicators

• atr: Average True Range
• rsi: Relative Strength Index
• bollinger: Bollinger Bands (mid, upper, lower)
• macd: MACD (macd, signal, histogram)

Custom Features

You can register your own features:

from priceframe.features import feature
import polars as pl

@feature("my_custom_indicator")
def _feat_custom(df: pl.DataFrame, spec) -> pl.DataFrame:
    """Custom indicator implementation."""
    window = spec.params.get('window', 10)
    col = spec.col('price', default='close')
    
    return df.with_columns(
        pl.col(col).rolling_mean(window).over('symbol').alias(spec.name)
    )

# Use it
spec = FeatureSpec(
    name='f_custom_10',
    func='my_custom_indicator',
    input_col='close',
    params={'window': 10}
)

Data Format

PriceFrame uses a canonical long-format schema:

Column	Type	Description
symbol	string	Instrument identifier
interval	string	Time interval (e.g., '1d', '1h')
ts	timestamp[ms, tz='UTC']	Timestamp
open	float64	Opening price
high	float64	Highest price
low	float64	Lowest price
close	float64	Closing price
volume	float64	Trading volume

Additional columns (features, metadata) are preserved.

Conversion Between Formats

# To pandas
df_pandas = pf.to_pandas(index='ts')  # or index='multi' for MultiIndex

# To Polars
df_polars = pf.to_polars()

# To PyArrow
table = pf.to_arrow()

# To Excel
pf.to_excel('output.xlsx')

# From close matrix
close_df = pd.DataFrame({
    'AAPL': [150, 151, 152],
    'MSFT': [300, 301, 302]
}, index=pd.date_range('2024-01-01', periods=3, tz='UTC'))

pf = PriceFrame.from_close_matrix(close_df, interval='1d')

Architecture

PriceFrame is built on three main components:

1. Core (priceframe.core): The PriceFrame class and data manipulation functions
2. Features (priceframe.features): Declarative feature system with registry
3. IO (priceframe.io): Data source connectors (Bloomberg, databases, etc.)

Design Principles

• Immutability: Most operations return new PriceFrame instances
• Type Safety: Automatic type casting and validation
• Performance: PyArrow backend for efficient columnar operations
• Flexibility: Support for both pandas and Polars workflows

Testing

Run the test suite:

# Install dev dependencies
pip install -e ".[dev]"

# Run tests
pytest

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Roadmap

• Additional data sources (Yahoo Finance, Binance, S3 storage, etc.)
• More technical indicators
• Backtesting framework integration (NautilusTrader)
• Performance optimizations
• Quick visualisation with mlpfinance

Requirements

• Python 3.9+
• pandas >= 2.3.3
• polars >= 1.35.2
• pyarrow >= 21.0.0

Optional:

• xbbg >= 0.7.7 (Bloomberg)
• blpapi >= 3.25.10.1 (Bloomberg)
• ccxt >= 4.5.20 (Crypto)
• duckdb >= 1.4.2 (Crypto storage)