In Round 1 we examined the prediction of the SP500 index with machine learning methods using TensorFlow in Python and H2O in R. Although both performed nearly identically in predicting prices over time, H2O in R was found to confer better loss protection in volatile conditions by a slim margin. Will this relationship hold true in the currency market. The present study will continue the series by examining TensorFlow in Python and H2O in R for predicting currency prices in Round 2. The results of the EURUSD currency pair will be analyzed in the pre, intra, and post pandemic period, and analysis characteristics between Rounds 1 and 2 compared. Differences between problem types, or financial markets, while seeking the identical outcome of profitability and conservation of principle will be described. We will evaluate the effectiveness of open source machine learning tools using workstation CPU, and examine the story expressed by the data, both the relationship between predictors and outcome, and the relationships between the explanatory variables themselves when constructing new machine learning models in a continuing variety of financial markets.