TY - GEN
T1 - Rank Prediction for Portfolio Management Using Artificial Neural Networks
AU - Bae, Jiyoon
AU - Sim, Ghudae
AU - Yun, Hyungbin
AU - Seok, Junhee
N1 - Funding Information:
This work was supported by the National Research Foundation of Korea grant (NRF-2017R1C1B2002850) and a grant from Mirae Asset Global Investments Co. Ltd. Correspondence should be addressed tojseok14@korea.ac.kr By using a modified version of a previously proposed approach using vocal tract length perturbation (VTLP) and a novel data augmentation approach based on stochastic feature mapping (SFM) in a speaker adaptive feature space, Cui showed that the proposed method can improve recognition performance by using both cross-entropy (CE) and state-level minimum Bayes risk (sMBR) as the cost function of ANN models. Hellstrom [6] proposed rank measure method taking into account a large number of securities and grades them according to the relative returns. Hellstrom showed that rank measure, besides being more related to a real trading situation, can be more predictable than the individual returns. As the rank prediction is easier than numeric prediction, in field of finance, to be specific, quantitative analysis, for overcoming existing algorithm, it is expected to be a useful method to apply results of rank prediction to predict profit.
Publisher Copyright:
© 2018 IEEE.
PY - 2018/8/14
Y1 - 2018/8/14
N2 - The rank of equities is often used to determine the investment portfolio instead of prices because ranking is in general believed to be robust. In this paper, we propose a rank prediction method for portfolio management using ANN. While an ANN requires a large dataset to train the model, the sample size is usually insufficient in stock market data. Therefore, the proposed method uses data augmentation and an ensemble ANN model. In the simulation study, the proposed method shows 13 percentage of performance improvement from the other methods to predict the profit rank of equities in South-East Asian market.
AB - The rank of equities is often used to determine the investment portfolio instead of prices because ranking is in general believed to be robust. In this paper, we propose a rank prediction method for portfolio management using ANN. While an ANN requires a large dataset to train the model, the sample size is usually insufficient in stock market data. Therefore, the proposed method uses data augmentation and an ensemble ANN model. In the simulation study, the proposed method shows 13 percentage of performance improvement from the other methods to predict the profit rank of equities in South-East Asian market.
KW - arfificial neural network
KW - portfolio management
KW - stock market prediction
UR - http://www.scopus.com/inward/record.url?scp=85052537937&partnerID=8YFLogxK
U2 - 10.1109/ICUFN.2018.8436983
DO - 10.1109/ICUFN.2018.8436983
M3 - Conference contribution
AN - SCOPUS:85052537937
SN - 9781538646465
T3 - International Conference on Ubiquitous and Future Networks, ICUFN
SP - 15
EP - 17
BT - ICUFN 2018 - 10th International Conference on Ubiquitous and Future Networks
PB - IEEE Computer Society
T2 - 10th International Conference on Ubiquitous and Future Networks, ICUFN 2018
Y2 - 3 July 2018 through 6 July 2018
ER -