Great article, thanks for sharing
I myself started some initiative to compare different models, doing it for around 10 days for LSTM, RNN, Arisma, RandomForest, LinearRegression and XGBoost models.
Article for LSTM in Medium - https://medium.com/python-in-plain-english/next-move-into-stock-prediction-lstm-networks-afb12e2fbf44
I also included in my daily comparison few metrics, to follow up on accuracy of the models - same metrics for all of them, applied on 5 different stocks. Strangely, LSTM wasn't the worst among models by an average success rate.