Possible follow - up questions

• Replace the LSTM with an RNN, and then with a GRU. Who's the best performer?
• Instead of predicting the closing price, try predicting also the high/low the day after. To do so, you can use the same features while training the model (or you can just use the closing price as input).
• Optimize the model for other stocks: is it better to have a generic model working for all the stocks or one specific for each stock?
• Tune the retraining. In the example, we predicted a full year with the model. Can you notice any improvement if you train the model once a month/week/day?
• If you have some financial experience, try building a simple trading simulator and feed it with the predictions. Starting the simulation with $100, will you gain or lose money after a year?