A collection of applied projects in time-series econometrics, volatility modeling, and Monte-Carlo methods for derivatives pricing — combining rigorous statistical modeling with real market data and written reports.
| Project | Topic | Methods | Stack |
|---|---|---|---|
| sales-forecasting | Time-series sales forecasting | ARIMA, SARIMA, SARIMAX, gradient boosting, naive baselines | Python |
| volatility-har-garch | Realized-volatility forecasting | HAR / HARQ / HARJ vs. GARCH family (GARCH, EGARCH, TGARCH, IGARCH, NAGARCH) | Python, R |
| monte-carlo-option-pricing | Derivatives pricing | Monte-Carlo simulation under a CEV process | Python |
Each project folder contains code, data (or a data link), generated plots, and a written report
(LaTeX → PDF). English READMEs are primary; the original Russian write-ups are kept as README.ru.md.
- Econometrics done properly — stationarity testing, model identification (ACF/PACF),
residual diagnostics, and out-of-sample evaluation (MAE / MSE / MAPE), not just
.fit(). - Model comparison — classical econometric models benchmarked head-to-head against gradient boosting and naive baselines on the same data.
- Real market data — Kaggle store-sales data and Moscow Exchange equity series.
Course projects, Vega Institute. Author: Maksim Kiryakin.