As part of a machine learning assignment at university we built a set of predictive models for five day ahead gold price. This needs some improvements with the use of time series econometrics and to be embedded in a useable dashboard but we are working on this as a longer term project.
Predicting Gold Prices
2.23MB ∙ PDF file
Within financial markets, predicting asset prices is an important topic with implications both for academia and in a commercial setting. This project aims to predict the close price of gold (XAU) five days into the future by using six key market variables and applying a variety of machine learning methods to find the best model to do so. The relevance of this prediction question comes from creating a set of models that will enable professional investors and academics to understand the relationships between market parameters and what they indicate about the future. Within a professional setting, accurate predictions would enable investors, financial analysts and institutions to make informed investment decisions to optimise their portfolios and mitigate market risks. Within the context of rules-based or algorithmic trading, a model which can reliably predict gold prices could be used to create a profitable trading strategy. For an academic, understanding the relations between variables and their ability to predict gold prices contributes to strengthening understanding of financial markets and macroeconomic dynamics.
Download
