ARIMAX Modelling for Brazilian Exchange Rate Forecasting
Budget: $30 – $250 USD
I am looking for a freelancer to build ARIMAX models to forecast EUR/BRL (eur_ptax_average, eur_ptax_end_of_month), GBP/BRL (gbp_ptax_end_of_month, gbp_ptax_average), USD/BRL (usd_ptax_average) and JPY/BRL (jpy_ptax_end_of_month, jpy_ptax_average), considering two types of monthly targets:
1. End-of-month exchange rates; and
2. Monthly average exchange rates.
The models should use the USD/BRL(usd_ptax_end_of_month) as exogenous variable. This variable is available in the attached Excel file dados_moedas.xlsx, in column D and highlighted in yellow.
The freelancer may use Python or EViews to develop the models. The forecast horizon should be determined by the number of future months available for the exogenous USD/BRL (usd_ptax_end_of_month).
The models must be economically interpretable, statistically well diagnosed, and tested against strong benchmark models, especially a random walk benchmark, given the well-known difficulty of forecasting exchange rates. End-of-month and monthly-average targets must be modeled and evaluated separately.
The final model should only be presented as useful if it demonstrates satisfactory out-of-sample performance, stable residual behavior, and plausible forecast paths. Residual diagnostics must be consistent with standard time-series econometric literature, including tests for autocorrelation, heteroskedasticity, normality, and model stability.
The final deliverables should include the estimated models, diagnostic tests, benchmark comparisons, out-of-sample performance metrics, forecast tables, and charts showing the projected paths for EUR/BRL, USD/BRL, GBP/BRL and JPY/BRL.
1. End-of-month exchange rates; and
2. Monthly average exchange rates.
The models should use the USD/BRL(usd_ptax_end_of_month) as exogenous variable. This variable is available in the attached Excel file dados_moedas.xlsx, in column D and highlighted in yellow.
The freelancer may use Python or EViews to develop the models. The forecast horizon should be determined by the number of future months available for the exogenous USD/BRL (usd_ptax_end_of_month).
The models must be economically interpretable, statistically well diagnosed, and tested against strong benchmark models, especially a random walk benchmark, given the well-known difficulty of forecasting exchange rates. End-of-month and monthly-average targets must be modeled and evaluated separately.
The final model should only be presented as useful if it demonstrates satisfactory out-of-sample performance, stable residual behavior, and plausible forecast paths. Residual diagnostics must be consistent with standard time-series econometric literature, including tests for autocorrelation, heteroskedasticity, normality, and model stability.
The final deliverables should include the estimated models, diagnostic tests, benchmark comparisons, out-of-sample performance metrics, forecast tables, and charts showing the projected paths for EUR/BRL, USD/BRL, GBP/BRL and JPY/BRL.