Alpha Discovery Neural Network based on Prior Knowledge

TL;DR

This paper introduces Alpha Discovery Neural Network (ADNN), a neural network model that leverages prior knowledge to automatically generate diverse financial indicators, outperforming genetic programming in feature construction for trading strategies.

Contribution

The paper presents a novel neural network architecture that uses domain knowledge, pre-training, and model pruning to efficiently construct diversified financial features, replacing traditional genetic programming methods.

Findings

ADNN constructs more informative and diversified features than GP.

Fully-connected and recurrent networks outperform CNN in extracting financial information.

Features from ADNN improve strategy revenue, Sharpe ratio, and reduce max draw-down.

Abstract

Genetic programming (GP) is the state-of-the-art in financial automated feature construction task. It employs reverse polish expression to represent features and then conducts the evolution process. However, with the development of deep learning, more powerful feature extraction tools are available. This paper proposes Alpha Discovery Neural Network (ADNN), a tailored neural network structure which can automatically construct diversified financial technical indicators based on prior knowledge. We mainly made three contributions. First, we use domain knowledge in quantitative trading to design the sampling rules and object function. Second, pre-training and model pruning has been used to replace genetic programming, because it can conduct more efficient evolution process. Third, the feature extractors in ADNN can be replaced by different feature extractors and produce different functions. The experiment results show that ADNN can construct more informative and diversified features than GP, which can effectively enriches the current factor pool. The fully-connected network and recurrent network are better at extracting information from the financial time series than the convolution neural network. In real practice, features constructed by ADNN can always improve multi-factor strategies' revenue, sharpe ratio, and max draw-down, compared with the investment strategies without these factors.