Actuarial Science: Optimising Trading Strategies by Deep Learning

 

Individuals often face various trading constraints in the financial market, such as restrictions on short-selling, no borrowing from outside, minimum capital requirement, and risk measure. It is challenging to find mathematical solutions to optimal trading strategies under these constraints. Since his previous work at the University of Waterloo, our researcher has been employing a deep-learning approach to construct a diversified portfolio by considering popular financial products like stocks, bonds, and life insurance, and then use a neural network to determine the ideal allocation of funds for each product. 

 

These strategies can help individuals outperform inflation, increase investment earnings, and manage mortality risk (death risk) under trading constraints. The result suggests that individuals tend to reduce their demand for life insurance when considering trading constraints. Moreover, the pioneering work can inspire more future research applying deep learning algorithms to portfolio management.

 

在金融市場中常常面臨各種交易限制。我們的研究人員 採用了深度學習方法，通過考慮股票、債券和人壽保險 等熱門金融產品，構建了一個多元化投資組合，然後利 用神經網絡確定了將資金投放在每個產品的理想配搭。