Financial Modelling in Python using several techniques like Deep Learning, etc.

Hello , I am a German Banker who is interested in Algorithmic Trading. With 5 years+ experience in Software like Tradestation or Multicharts I am experienced in Financial Modelling in those Finance applications. When it comes to Python I am still in the learning phase. This is one of the reasons I am reaching out to you. When it comes to Python, it is often crucial which library version is installed, if the module is installed at all, etc. So my first question is: For Code execution should a Cloud environment be used where we can both access? I came across the following idea. (As you might have a Math background you might be able to understand the logic behind it). Here is the basic idea, make sure you read it. [login to view URL] Short explanation and obstacles of the code: 1. For the timeseries prediction a stock quote is downloaded with two columns: Date and Close Price. In this case Goldman Sachs stock was downloaded (any other can also be used for testing). 2. The data is split in training and test data 3. Correlated assets: This is a little bit vague in the article. It says 72 additional data sets on Daily basis are also downloaded and used, not exactly indicating which ones (we can discuss the symbols to be used here). 4. Technical indidcators are added to the dataset. 5. Fundamental data is added to the dataset: Sentiment analysis with BERT module to calssify the companies headlines. 6. Fourier transforms for the sine wave. 7. ARIMA is used as a technique for predicting time series data. 8. After having added all the features above we now have 112 features (columns): Feature importance test with XGBoost. 9. Autoencoder: more features are generated  not evaluated further 10. The 112 features mentioned for the correlated assets, technical indicators, fundamentl analysis, Fourier, and Arima are evaluated for usefulnes Eigen portfolio with PCA. 84 are left over. 11. GAN and the Reinforcement learning parts are missing here. Are you aware of code using GAN and can it be applied here? 12. LSTM + Learning Rate is applied 13. CNN Architecture is applied 14. Bayesian optimization 15. Result after different epochs of training Tasks: 1. Carefully read the article and apply the code (missing parts might be added at a later stage). Can the predicted values be replicated? 2. How does the predicted number of future days affect the quality of the prediction? 3. Can we find ways to easily amend the datafeed (automate) for other symbols and also customize the correlated assets? Once these tasks are accomplished we can think about further tasks. 4. In Python are modules/libraries like Backtrader/Zipline to validate the backtest with KPIs. A backtest of the strategy should be implemented. We can happily discuss the code and the mentioned steps for further clarification. As I plan to build my Algorithmic business in Germany and we can succesfully accomplish this project I am not averse to offer you future projects, even a stakes oft he company as the main programmer.