Smart Portfolio selection

Smart Portfolio selection

Problem being addressed

Portfolio Selection is an important real-world financial task and has attracted extensive attention in artificial intelligence communities. This task​,​ however​,​ has two main difficulties: the non-stationary price series with complex asset correlations​,​ and fact that the practicality principle in financial markets requires controlling both transaction and risk costs.

Solution

A cost-sensitive portfolio policy network to address it via reinforcement learning. To extract meaningful features, a novel two-stream network architecture is developed to capture both price sequential information and asset correlation information. To control both transaction and risk costs, the researchers developed a new cost-sensitive reward function.

Advantages of this solution

The suggested network makes more profitable decisions. By exploiting reinforcement learning to optimize the reward function, the proposed network is able to maximize the accumulated return while controlling both costs. Extensive experiments on real-world datasets demonstrate the effectiveness and superiority of the proposed method in terms of profitability, cost-sensitivity and representation abilities.

Solution originally applied in these industries

finance

Financial Sector

Possible New Application of the Work

data:image/png;base64,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

Media

Social media can already be a valuable source of data to make all sorts of predictions, including the political and financial ones. The suggested algorithm can further explore the correlation between social text information and price sequential information.

Author of original research described in this blitzcard:

question-mark-icon

Name of the author who conducted the original research that this blitzcard is based on.

Source URL: #############check-icon


search-iconBrowse all blitzcards