In one of our projects, we designed an intelligent asset allocation system that utilized Deep Learning and Modern Portfolio Theory.
The task was to implement an investment strategy that could adapt to rapid changes in the market environment.
The base AI model was responsible for predicting asset returns based on historical data. This was accomplished by implementing Long Short-Term Memory Units, which are a sophisticated generalization of a Recurrent Neural Network. This particular architecture can store information for multiple timesteps, which is made possible by a Memory Cell. This property enables the model to learn long and complicated temporal patterns in data. As a result, we were able to predict the asset’s future returns, as well as the uncertainty of our estimates using a novel technique called Variational Dropout.
In order to strengthen our predictions, we used a wealth of market data, such as currencies, indices, etc. in our model, in addition to the historical returns of relevant assets. This resulted in over 400 features we used to make final predictions. Of course, many of these features were correlated. This problem was mitigated by Principal Component Analysis (PCA), which reduces the dimensionality of the problem and decorrelates features.
We then used the predictions of return and risk (uncertainty) for all the assets as inputs to a Mean-Variance Optimization algorithm, which uses a quadratic solver to minimize risk for a given return. This method determines the allocation of assets, which is diverse and ensures the lowest possible level of risk, given the returns’ predictions.
Combining these models created an investment strategy which generated an 8% annualized return, which was 23% higher than any other benchmark strategy tested over a two year period.