Breakthrough Forecasting Techniques: Evaluating Copula Models For Canadian Dollar And Major South Asian Currencies

Abstract

Owing to the advancement in financial globalization, forecasting on exchange rates has become relevant to investors, businesspersons, and policymakers. Standard paradigms such as time series analysis can find it rather challenging to capture the inherent non-linear features characteristic of exchange rate Data. Copula models present a very reliable alternative to examine the dependence structure of several markers because they allow a general way to represent the distribution of several variables without the need for normal distribution and linear functional forms. Thus, the current study examines the use of copula models to predict the exchange rates of CAD/PKR, CAD/INR, and CAD/LKR.  While specifying copula models, it is found that these models can forecast the extreme movements of the market and the dependencies between the spot and forward exchange rates more effectively than the others. The analysis of the results reveals that copula models provide higher predictive performance rather than more conventional techniques by following metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). For the second aspect, the density surfaces of the copula density prove the existence of a positive and high correlation between the spot and forward rates, hence the ability of the model to increase the reliability of the exchange rates forecasts. Consequently, the present work highlights the importance of including the copula structures in the context of finance with specific regard to situations characterized by high volatility. Therefore, the copula models enhance quantitative risk management and exploit dependencies in the currency market as an opportunity for the financial sector.