Skip to content

Latest commit

 

History

History
32 lines (18 loc) · 1.4 KB

README.md

File metadata and controls

32 lines (18 loc) · 1.4 KB

orthogonal-bandits

Orthogonal Bandit Portfolios

Usage

ticker = ['VOW.DE','BA', 'AMD', 'AAPL','GME','CVGW','CAMP','WSCI','LNDC','WOR'] #<- ticker lists
start_date=[2007, 9, 1] #<- starting history date
end_date=[2009,12,9] #<- ending history date tau = 300 #<- Length of the rolling window (starting fro, start date and shifting by +1 at each inference step)
Nsign = 4 #<- number of significant portfolios. To be tuned.

Inference window is [start_date+tau+1:end_date].

Cumulative Wealth (CW) comparison among OBP, OBP w/o shortsale and Equally Weighted (EW) portfolios.
alt text

Theta mixing among relevant and irrelevant orthogonal portfolios (theta=1 all irrelevant, theta=0 all relevant)
alt text

Invested percentage for OBP. Negative values correspond to short sale balancing values larger than 100%.
alt text

Invested percentage for OBP without short sale possibility. Minimum % is clipped to 0 and all other weights are normalized consequently.
alt text

Inspired by "Portfolio Choices with Orthogonal Bandit Learning" W. Shen et al. 2015.