Testing Oil and Gold Inputs for USDZAR Forecasting

Summary

This article evaluates whether oil and gold prices improve forecasts of USDZAR compared with the pair’s own OHLC data. It describes exporting price histories, reversing their order into chronological sequence, aligning observations by timestamp, and predicting the USDZAR close at a stated future horizon. The authors compare ordinary USDZAR inputs with a combined oil and gold feature set using several regression models and time-series cross-validation. The reported analysis finds a stronger correlation between oil and USDZAR than between gold and USDZAR, while scatter plots show no clear separation. The lowest error came from linear regression using USDZAR OHLC data; among the commodity-input models, a tuned K-nearest-neighbors regressor performed best. The article also describes holding out validation data to compare a customized model with a default one, then deploying a selected model in an MQL5 expert advisor. Results are specific to the tested data and symbol: correlation alone does not establish a causal or stable relationship, and the authors caution that other instrument baskets may perform differently.

Key ideas

• The study compares USDZAR OHLC predictors with oil and gold price inputs for forecasting the pair’s future close.
• Oil showed a stronger reported correlation with USDZAR than gold, but the visual analysis did not reveal a strong relationship.
• Time-series cross-validation favored linear regression on USDZAR OHLC data, while K-nearest neighbors led among the commodity-input models.
• The article describes tuning and validation against a default model before integrating a customized model into an MQL5 expert advisor.
• The findings are limited to the tested data and do not establish that commodity prices causally or reliably predict USDZAR.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.