fractime 0.1.0

Python library for advanced time series forecasting using fractal geometry and chaos theory

FracTime

Fractal-based time series forecasting with ensemble methods and rigorous backtesting.

Installation

pip install fractime

All dependencies (ARIMA, GARCH, Prophet, PyMC, XGBoost) are included.

Quick Start

import fractime as ft
import numpy as np

# Load or create data
prices = np.random.randn(500).cumsum() + 100

# Analyze fractal properties
analyzer = ft.FractalAnalyzer()
hurst = analyzer.compute_hurst(prices)
print(f"Hurst: {hurst:.3f} ({'trending' if hurst > 0.5 else 'mean-reverting'})")

# Forecast
forecaster = ft.FractalForecaster()
forecaster.fit(prices)
result = forecaster.predict(n_steps=30)

print(f"Forecast: {result['forecast'][-1]:.2f}")
print(f"95% CI: [{result['lower'][-1]:.2f}, {result['upper'][-1]:.2f}]")

# Visualize
fig = ft.plot_forecast(prices, result['forecast'], result['paths'])
fig.show()

Features

• Fractal Forecasting: Hurst exponent, fractal dimension, long-term memory
• Baseline Models: ARIMA, ETS, GARCH, Prophet, VAR, LSTM
• Ensemble Methods: Stacking and boosting
• Backtesting: Walk-forward validation with comprehensive metrics
• Model Selection: Automatic selection with statistical testing

Documentation

Full documentation: https://wayy-research.github.io/fracTime

• Installation Guide
• Quick Start Tutorial
• API Reference
• Examples

License

MIT