← Back to projects

Foreign Exchange Rate Forecasting

Built a forecasting pipeline for the EUR/USD currency pair using statistical baselines and deep learning. Combined classical time series analysis with recurrent and convolutional networks to surface medium-term trading signals for a graduate analytics project.

Timeline
Jan 2024 – Apr 2024
Role
Graduate Researcher
Stack
Python, R, pandas, statsmodels, TensorFlow, Keras
Dataset
6,119 historical EUR/USD observations (daily)

Approach

  • Engineered features from macroeconomic indicators and technical signals (momentum, volatility, moving averages) to enhance model sensitivity.
  • Benchmarked ARIMA/SARIMA models before progressing to stacked RNN and CNN architectures for capturing both long- and short-term patterns.
  • Tuned hyperparameters with rolling-window cross validation to align training with real-world deployment constraints.
  • Created dashboards that visualized forecasts, confidence intervals, and error diagnostics for presentation to faculty.

Highlights

  • Improved directional accuracy over statistical baselines and produced sharable notebooks to support scenario analysis.
  • Delivered a narrative report translating complex model outputs into actionable insights for technical trading recommendations.
  • Documented an extensible roadmap for incorporating exogenous features and live data feeds in future iterations.