About StochX

StochX is an independent academic and software project developed around two connected goals:

  1. make stochastic-process mathematics executable and testable;
  2. provide a lightweight, reproducible Python workflow for practical time-series analysis and forecasting.

Main applied focus

The time-series layer is designed for students, researchers and analysts who are comfortable with the EViews methodology and want to move the same style of analysis into Python.

The project therefore keeps familiar concepts such as workfiles, samples, lags, differences, equations, ADF, correlograms, ARMA errors, model diagnostics and forecasting while exposing them as scriptable Python objects.

Academic foundation

The stochastic-process layer grew out of graduate-level coursework in stochastic modeling and forecasting and covers Markov chains, Poisson processes, continuous-time Markov chains, birth-death processes, conditional expectation and martingale concepts.

Design philosophy

StochX prefers:

  • explicit mathematical assumptions;
  • small composable objects;
  • deterministic numerical conventions;
  • reproducible examples;
  • report-oriented results;
  • clear separation between theory, user workflow and API reference.

Reference

Mazliak, L., Priouret, P., & Baldi, P. Martingales et chaînes de Markov. 1998.