JDE65 / AssetAnalysis

Asset analysis : downloading assets and correlated assets for analysis with neural nets

AssetAnalysis - WIP

Asset analysis : downloading Assets infos and correlated assets for analysis The schema always follows the structure presented herewith.

The repository includes (and/or will include) various codes to :

0. Load the libraries

1. Ask for inputs : hyper-parameters & parameters

2. Connect to DBs and load the data

• download stock data (date, close, open, high, volume, daily traded volume, ...)
• dowload financial data of "correlated assets" (major indices, ST and LT interest rates, commodities, ...)
• enrich data with technical indicators (MA & EMA, MACD, Bollinger band, CCI, EMV, ATR, ADX, RSI, MOM)
• enrich data with price from assets of the same class
• enrich data with correlated assets (FX, rate, commodities, macro-economic data, ...)
• transform series for detailed analysis (Fourier, ARIMA, ...)

3. Organise the data into matrixes X and y

• Create the Dataset
• Enrich the Dataset
• Organize matrix X
• Organize matrix y
• Clean X & y for invalid data
• Normalize X
• Separate into train & test sets

4. Organize the model architecture

• Dense NN
• LSTM NN
• Conv1D NN
• ConvLSTM 1D NN
• Bayesian LSTM -...

5. Compile + Train + Test the model

6. Perform scenario analysis for these NN

• Compute the expected results of the strategy
• Analyze number of deals, IRR, average investment, Sharpe & Sortino ratios
• Plot the marked-to-market
• Plot average daily return

... TBC