I built a Neural Network model (using Python and TensorFlow) for forecasting stock prices in 2016 and used the algorithm in my hedge fund in 2017. I made a video which provides a step-by-step walk through for how to create, train, and evaluate the model for the the purpose of generating buy and sell signals for financial securities.

The methodology should be very useful for anyone looking to build something like this.

This example utilizes a four layer neural network for trading volatility ETPs but can be easily modified to include other securities and metrics as predictor inputs.