Multiple strategies applied to trading such as regression, classification and trend following.
Trend following strategy applied in trading, with a for loop looking for the best windows time on cryptocurrencies returns.
Linear regression applied as a classification approach, due the output was analyzed as positive or negative return, not integer value.
Logistic regression applied shifting a number of assets returns to classify the following returns as positve or negatives, not measured in distance only in direction.
Random forest classifier applied for a similar approach than logistic regression.
All data was requested from kraken cryptocurrency exchange API, feel free to run notebook and see current results.