Next-day KOSPI closing price prediction using regression models.
| Part 1 (OHLCV only) | Part 2 (+ Technical Indicators) | |
|---|---|---|
| Best Model | Ridge Regression | Linear Regression |
| R² | 0.8315 | 0.8971 |
| RMSE | 33.41 | 24.23 |
| MAE | 26.51 | 18.22 |
R² improved from 0.83 → 0.90 by adding technical indicators.
Part 1 — OHLCV lag features (1, 2, 3, 5 days)
Part 2 — Part 1 + MA5/20/60, RSI, Bollinger Bands, Momentum, Daily Return, Volatility
Linear Regression, Ridge, Lasso, ElasticNet, Random Forest, Gradient Boosting, XGBoost
TimeSeriesSplit cross-validation (prevents data leakage)
Python, Scikit-learn, XGBoost, Pandas, NumPy, Matplotlib