Deep Learning Asset Allocation

Summary (Eng)

Asset allocation refers to the portfolio construction strategy by appropriately allocating proportions to available asset classes. With asset allocation, investors can significantly reduce the overall risk of portfolio by investing in assets ranging from high risk assets such as stocks, to assets with low risks such as bonds and cash. Increasing the rate of return is crucial for an investment, but reducing risk is equally important to achieve a steady return. Our purpose of asset allocation using deep learning is to provide the model that delivers better returns with similar risks compared to existing well-known benchmarks.

Traditional asset allocation investment is widely adapted by many asset managers in various forms, and individuals also utilizes these strategy without difficulty. The traditional asset allocation method assumes the continuity of the historical distribution that the past data distribution will be similar in the future, so it is hard to consider to be realistic with our view in the market. In addition, even if tactical asset allocation such as Momentum strategy and Long-Term Reversal strategy for portfolio performance is implemented, we cannot be sure whether specific parameters are optimal. To solve these problems and to build better strategies, we use artificial intelligence technology rather than human trial-and-error. With A.I. driven process, it is possible to minimize the time spent to optimize well-known quant strategies and deliver better risk-adjusted return. When examining the performance of asset allocation through deep learning with simple empirical analysis, it showed better performance than the not only traditional asset allocation strategy, the 60/40 portfolio and the risk-parity strategy but Machine Learning Methodologies.

We effectively establish investment strategy and execute process using financial data through A.I Technologies, even achieve better performance of strategies. The Asset Allocation strategy utilizing Deep Learning is expected to lead to a chance in the financial market, as there is probability for simplifying the process and improving performance as the technology develops.