Examining the feasibility of quant methods in finance (a proof of concept).
Initialize a python virtual environment and pip install requirements (keras, transformers, numpy, pandas, csv).
- LSTM - run
python testing.py
from the root directory - BERT -
cd
intosentiment/
, then runpython emotion.py
Testing various RNN models to find strengths/weaknesses of each as it applies in micro & macroeconomic theory.
- LSTM: univariate (vanilla, stacked, bidirectional); multivariate; multiparallel
- NLP: BERT (Bidirectional Encoder Representations from Transformers)
Data in csv format, preprocessed to calculate log returns.
- S&P 500: weekly intervals for 20 years
- VIX index to measure volatility