Overview
At Datraxa, we built a complete time series forecasting pipeline to predict USD/TRY exchange rates using real financial data from Yahoo Finance.
The project combines data preprocessing, statistical testing, and predictive modeling to generate accurate and reliable forecasts using ARIMA and SARIMA models.
Our workflow demonstrates how statistical forecasting can reveal hidden market patterns, guide financial decision-making, and improve exchange rate prediction accuracy for traders, analysts, and financial institutions.
🔍 Key Features
- Exploratory Data Analysis (EDA):
Performed STL decomposition, volatility tracking, moving averages, and daily fluctuation analysis. - Stationarity & Seasonality Testing:
Applied ADF, KPSS, Hodrick–Prescott filter, and Kruskal–Wallis tests to ensure model reliability. - Predictive Modeling:
Built and tuned ARIMA and SARIMA models using cross-validation and seasonal order optimization. - Model Evaluation:
Compared MAE, RMSE, and MAPE metrics with detailed residual diagnostics (Ljung–Box test). - Backtesting Framework:
Implemented rolling-window validation to assess long-term model stability and forecast robustness.
⚙️ Tech Stack
Python · Pandas · Statsmodels · Matplotlib · TimeSeriesSplit · ARIMA · SARIMA
💾 Dataset
- Source: Yahoo Finance – USD/TRY
- Frequency: Daily
- Features: Open, Close, High, Low, Adjusted Close, Volume
📊 Visual Highlights
- STL decomposition of the USD/TRY series
- ACF/PACF plots for ARIMA tuning
- Forecast comparison graphs (ARIMA vs. SARIMA)
- Moving average and volatility trend visuals
- Ljung–Box residual diagnostics plots
This project highlights Datraxa’s expertise in financial analytics, showcasing how advanced time series modeling transforms complex currency data into clear, actionable insights for smarter economic forecasting and decision-making.
💡 Insights
- ARIMA performed best under stable market conditions with low volatility.
- SARIMA effectively modeled cyclical and seasonal behaviors but required fine-tuned parameters.
- The pipeline can be adapted for other currencies, stock data, or macroeconomic indicators, serving as a reliable forecasting blueprint for real-world financial analytics.
🧭 Use Cases
- Currency exchange rate forecasting
- Financial planning and risk management
- Macro-economic scenario modeling
- Integration with algorithmic trading systems
Datraxa — Turning data into foresight with intelligent forecasting systems.