From your requirements, it's similar to my previous project.
This is project to predict stock marketing price. I analyzed which fields are needed for stock marketing and collected that. It have learnt this data and predict based on current data and trained model.
My Key POINTS for your idea: Rich experience in Python, stock marketing, LSTM/GRU.
Thank for my chance.
For the high accuracy, first we need to determine which fields affects to stock marketing. Second, we need to build/select correct model, in general, if we have enough training data, we can build large model. I prefer LSTM or GRU. Third, we can train the model and it is needed good find-tunning. It involves selection of initial value<->activation function, dropout rate, regularization(L1, L2, L1-L2). Fourth, test the model.