Regret Bounds for Online Portfolio Selection with a Cardinality Constraint

Authors: Shinji Ito, Daisuke Hatano, Hanna Sumita, Akihiro Yabe, Takuro Fukunaga, Naonori Kakimura, Ken-Ichi Kawarabayashi

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We show the empirical performance of our algorithms through experiments over synthetic and realworld data. In this section, we consider the online portfolio selection problem with S = S1.
Researcher Affiliation Collaboration Shinji Ito NEC Corporation Daisuke Hatano RIKEN AIP Hanna Sumita Tokyo Metropolitan University Akihiro Yabe NEC Corporation Takuro Fukunaga RIKEN AIP, JST PRESTO Naonori Kakimura Keio University Ken-ichi Kawarabayashi National Institute of Informatics
Pseudocode Yes Algorithm 1 An algorithm for the full-feedback setting. and Algorithm 2 An algorithm for the bandit-feedback setting.
Open Source Code No The paper does not provide an explicit statement about releasing its source code or a link to a code repository for the methodology described.
Open Datasets Yes The first is based on crypto coin historical data2, including dates and price data for 19 crypto coins. From this data, we select 7 crypto coins, each having 929 prices, and obtain price relatives rti of coin i at time t by (pti/pt 1,i) 1, where pti indicates the price of coin i at time t. Thus, d = 7 and T = 928 in this instance. The other instance is based on S&P 500 stock data3, including dates and price data for 505 companies. From this data, we choose d = 470 companies, each having 1259 stock prices, and compute T = 1258 price relatives for each company in the same way. 2https://www.kaggle.com/sudalairajkumar/cryptocurrencypricehistory 3https://www.kaggle.com/camnugent/sandp500
Dataset Splits No The paper describes how real-world data is processed and selected (e.g., 'select 7 crypto coins', 'choose d = 470 companies'), but it does not specify exact split percentages, absolute sample counts, or refer to predefined splits for training, validation, and testing needed to reproduce data partitioning.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper discusses algorithms and methods used, but does not provide specific ancillary software details like library or solver names with version numbers needed to replicate the experiment.
Experiment Setup Yes In particular, setting η = 1 C4 min 1, q and β = 1, we obtain E[RT ] = O p T log |S| + k log T + log |S| . (5) (for Algorithm 1, Theorem 2) and Setting γ = min 1, q Tk|S| log(1+T ) , η = γ C4|S| min n 1, q log |S| k log(1+T ) o and β = C3C5, we obtain E[RT ] = O p T|S|k log T + |S| p k log |S| log T + |S|k . (for Algorithm 2, Theorem 3). And We set parameters η according to Theorem 2 for Algorithm 1 and MWU_disc, and η and γ according to Theorem 3 for Algorithm 2, Exp3_disc, and Exp3_cont.