macrovolpy 0.1.0

Deep statistical forecasting and volatility learning for macro-financial time series.

MacroVolPy

MacroVolPy is a Python package for deep statistical forecasting and volatility learning in macro-financial time series.

It provides tools for:

• simulated macro-financial uncertainty data
• time-series transformation
• descriptive diagnostics
• CNN forecasting
• LSTM forecasting
• conditional mean-variance volatility learning
• Student-t likelihood
• prediction intervals
• ARIMA and GARCH benchmarks
• rolling-origin validation
• Differential Evolution search

Installation

pip install macrovolpy

For deep learning models:

pip install macrovolpy[deep]

For econometric benchmarks:

pip install macrovolpy[econometrics]

For development:

pip install macrovolpy[dev]

Quick start

from macrovolpy import simulate_macro_series
from macrovolpy import DeepStatForecaster

data = simulate_macro_series(n=360)
y = data['change'].values

model = DeepStatForecaster(
    model='cnn',
    optimizer='adam',
    lookback=12,
)

model.fit(y)

forecast = model.predict(horizon=12)

print(forecast)

Volatility intervals

from macrovolpy import VolatilityForecaster

vol_model = VolatilityForecaster(lookback=12)

vol_model.fit(y)

intervals = vol_model.prediction_intervals(
    horizon=12,
    levels=(0.80, 0.95),
)

print(intervals)

Citation

This package is based on the methodology introduced in:

Alakkari, K., Abotaleb, M., El-kenawy, E. M., and Mishra, P.

Advanced Deep Statistical Learning Approach for Forecasting Global Economic Policy Uncertainty and Volatility.

Computational Economics.

DOI: 10.1007/s10614-026-11350-7