Predicting financial markets has become increasingly relevant with the significant growth in financial risk levels and uncertainties. This paper aims to forecast three major cryptocurrencies (Bitcoin, Ethereum, and Litecoin) through commodity submarkets during the period 2015--2023, covering the COVID-19 crisis and Russian-Ukrainian crisis periods. In a forecasting target based on long short-term memory (LSTM) and support vector machine (SVM) covering the period from 2015 to 2023, the chapter's findings confirm the robustness of the deep learning model, proving the forecasting power of commodities and revealing the weight of the degree of compartmentalization and correlation between the two markets on the forecasting quality. It highlights corn as the best forecaster, noting that information disclosed from cryptocurrency-connected markets has better forecasting quality than internal past information. From another perspective, estimations confirm that the framework of the Ukrainian Russian crisis has better predictive power for cryptocurrencies than does the COVID-19 framework.