- R version 3.6.2
- Python version 3.8.2
The correct prediction of any asset is almost impossible, however, one must try it.
- Making use of google trends data in words such as [Bitcoin, Blockchain, debt, dollar, ...].
- Getting Bitcoin data such as open_time, close_price.
- Getting the returns from close_price.
- Merging both data bases to built the training and testing bases.
- Comparing multiples models to get the best option.
Best model with minimum number of variables.
glm(y ~ banking.system + bce + bill.gates + bitcoin.alert +
bitcoin.bubble + bitcoin.dump + bitcoin.hack + bitcoin.whales +
bitcoin + blockchain + coingecko + cryptonews + debt.jubilee +
financial.independence + housing.market + jerome.powell +
microsoft + ray.dalio + real.economy + real.estate + satoshi.nakamoto +
vitalik.buterin + volatility + amazon + apple + bitcoin.twitter +
ethereum + smart.contracts + sec + facebook + fed,
data = db_training, family = "binomial")