1: \begin{abstract}

2:     

3:     In online portfolio optimization the investor makes decisions based on new, continuously incoming information on financial assets (typically their prices).

4:     In our study we consider a learning algorithm, namely the Kiefer--Wolfowitz version of the Stochastic Gradient method,

5:     that converges to the log-optimal solution in the threshold-type, buy-and-sell strategy class.

6:     

7:     The systematic study of this method is novel in the field of portfolio optimization; we aim to establish the theory and practice of Stochastic Gradient algorithm used on parametrized trading strategies.

8:     

9:     We demonstrate on a wide variety of stock price dynamics (e.g. with stochastic volatility and long-memory) 

10:     that there is an optimal threshold type strategy which can be learned. 

11:     Subsequently, we numerically show the convergence of the algorithm.

12:     Furthermore, we deal with the typically problematic question of how to choose the hyperparameters 

13:     (the parameters of the algorithm and not the dynamics of the prices) 

14:     without knowing anything about the price other than a small sample.

15:    

16: \end{abstract}

17: