Kernel-Based Aggregating Learning System for Online Portfolio Optimization

Algorithm 3

Solving KAL system.

Input: Given parameters , , ϵ, asset prices in the recent time window and , the current portfolio .
(1)	Calculate by (5), the indicator information by (7).
(2)	Calculate by (9), and by (11), price prediction by (12).
(3)	Calculate the final price relative prediction by (14) and the adjust matrix by (19).
(4)	if then
(5)
(6)	else
(7)
(8)	end if
(9)	Optimization: .
(10)	Projection: .
Output: The next portfolio .