Using machine learning to predict clean energy stock prices: How important are market volatility and economic policy uncertainty?

The disruptive impacts of climate change have created an urgent need to transition to a low carbon economy and an important part of this transition is an increase in the usage of clean energy. The greater adoption of clean energy is creating new opportunities for clean energy equity investing. The existing literature mostly focuses on the dynamic relationship between clean energy equities, oil prices, technology stock prices, and other important macroeconomic variables like market volatility and economic policy uncertainty. However, there is a shortage of literature on forecasting clean energy stock prices. Forecasting clean energy equity prices is important for making investment decisions. This paper uses machine learning methods to predict the direction of clean energy stock prices. The analysis reveals that random forests, extremely randomized trees, stochastic gradient boosting, and support vector machine have higher prediction accuracy than Lasso or Naïve Bayes. For forecasts in the 10-day to 20-day range, random forests, extremely randomized trees, stochastic gradient boosting, and support vector machine achieve prediction accuracies greater than 85 %. In some cases, prediction accuracy reaches 90%. Lasso prediction accuracy is higher than Naïve Bayes but never greater than 65 %. The MA200, MA50, and WAD technical indicators are, on average, the features most important for predicting clean energy stock price direction. Of the non-technical indicators, VIX and OVX are consistently ranked high in importance. In most cases, EPU is not one of the most important features, Of the forecasting methods considered, extremely randomized trees are very impressive due to high accuracy and short computational time.