Gold price prediction using support vector regression and random forest regression

Gold is a valuable asset that is frequently bought and sold by investors and portfolio managers. It is considered a safe-haven investment because it helps protect against market downturns, maintains its value, and provides liquidity during uncertain times. However, gold prices can fluctuate frequently, making it important to predict its price in order to maximize profits and minimize losses. The study employs two methods Support Vector Regression (SVR) and Random Forest Regression using daily price data of various commodities from 2020 to 2022. The selected commodities include Gold Spot in US Dollars to represent gold price in circulation, Bloomberg WTI Crude Oil for oil price, US Dollar Index as an indicator of exchange rates, and S&P 500 to represent stock market values. The goal is to estimate the gold’s closing price on the next trading day. Three sets of data are analyzed: the first set looks at the closing prices from several days prior relative to the asset itself [1]; the second set examines the opening, closing, lowest, and highest prices of the asset from the previous day [2]; and the third set considers the opening, closing, lowest, and highest prices of other commodities from the preceding day [3]. The models are evaluated using three metrics: MAPE (Mean Absolute Percentage Error), MAE (Mean Absolute Error), and R-squared. The results show that SVR provides the most accurate predictions, with the price of gold at the close of trading from multiple days ago being the most important factor in predicting future gold prices.