Anticipating Cryptocurrency Prices Using Machine Learning

Figure 9

Method 1: parameters optimisation. The sliding window (a, d), the training window (b, e), and the number of currencies (c, f) chosen over time under the geometric mean (a, b, c) and the Sharpe ratio optimisation (d, e, f). Analyses are performed considering prices in BTC.

(a)

(b)

(c)

(d)

(e)

(f)