In this project, we evaluate the performance of bayesian Vector Auto Regressive models in the context of Indian exchange rate forecasting. Various priors are compared.
Presented this work as a talk titled, "Bayesian forecasting of Rupee/Dollar exchange rate: Does Minnesota prior matter?" at the International Conference On Financial Markets And Corporate Finance held in IIT Kharagpur, India.
A detailed explanation of the approach and discussion of the results can be found in the report.
I have also explored the time aware deep learning models like LSTMs for exchange rate forecasting. The project can be found in this repository.
- Report: Writeup of the details of the project
- data.csv: Contains data used in forecasting. Obtained from RBI's website
- bayes_var: Main matlab file. Compares and evaluates the various models
- mlag2.m: Helper file