The team also ran intial model and presented the analysis
Idea for mitigating overfitting is good
Improvement areas
5 explanatory variables might be low for predicting the bitcoin prices. The team can include additional predictors.
The team can experiment with different periods for different terms in Auto-regression
From a practical standpoint, I am not sure if just knowing +ve or -ve return is useful. As an investor, I would like to compare the returns across different alternatives. Hence, the % return from bitcon is important.
Improvement areas