StochX

A practical time-series library for Python

StochX is a lightweight scientific Python library with two clear layers:

• Time Series — the main applied layer for econometrics, forecasting and EViews-style analysis.
• Stochastic Processes — a separate mathematical layer for probability and stochastic-process coursework.

The time-series documentation is the best place to start for analysts who are moving an existing EViews workflow into Python.

What makes the time-series layer different?

StochX keeps the workflow familiar:

workfile → sample → series expressions → stationarity
        → ACF/PACF → model estimation → diagnostics
        → model selection → forecast → report

The implementation uses Python-native objects, but the vocabulary is intentionally recognizable: Workfile, C, @TREND, D(Y), DLOG(Y), Y(-1), wf.ls(...), ADF, AR/MA terms, correlograms, result tables and interpretation helpers.

The project does not claim to be a universal EViews clone. Instead, it makes the same statistical workflow easy to reproduce in Python, and it records exact numerical conventions in regression tests where parity has been benchmarked.

Start with the guides

Time Series

Read the Time Series User Guide first. It covers the complete applied workflow from data preparation to forecasting, plus the state-space/Kalman extension.

Stochastic Processes

Use the Stochastic Processes User Guide for discrete-time Markov chains, Poisson processes, CTMCs, birth-death processes, probability objects, martingales and simulation.

Course Material

The Course Material section separates mathematical explanations, notation and assumptions from the software API.

Package / API

The Time-series API map and Stochastic-process API map explain the public Python surface. The individual API pages contain the exact object-level reference.

Installation

python -m pip install stochx

Minimal time-series example

from stochx.timeseries import Workfile, adf

wf = Workfile.from_csv("macro.csv", date_column="DATE", frequency="M")
wf.set_sample("2010-01-01 2024-12-01")

print(wf.info())
print(wf.eval("GDP(-1)"))
print(wf.generate("DGDP", "D(GDP)").summary())

eq = wf.ls("GDP C CONS CONS(-1)", name="EQ01")
print(eq.summary())

print(adf(wf["GDP"], regression="c", lags=1, autolag=None).summary())

Quality

The package is tested across Python 3.10, 3.11 and 3.12. The repository also runs all examples and validates documentation builds and distribution artifacts.

The current stable package version is 0.3.0.