The present study aims to outline a general framework by examining various machine learning algorithms in financial time series forecasting. The study provides information about widely used tree-based Random Forests (RF), Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), Light Gradient Boosting Machine (LightGBM), and Natural Gradient Boosting (NGBoost) models in financial time series forecasting. Then, the SHAP approach is explained, enhancing the interpretability of machine learning models by measuring the contribution of input variables. Lastly, financial time series predictions conducted using machine learning algorithms are discussed through stock, bond, and commodity markets, demonstrating forecasting performance and effectiveness of these models. Machine learning models perform well on the predictions of commodity prices such as crude oil, natural gas, gold, agricultural products, as well as stock and bond markets. In addition, these models are proposed to be effective in the prediction of green energy stock and green bond markets, and their importance is emphasized for investment decisions, policy formation, and sustainable energy transformation. In conclusion, novel machine learning models exhibit superior performance in financial time series forecasting, and it is crucial with respect to offering advantages such as prediction accuracy, interpretability, and capturing non-linear patterns.