Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The cross-sectional stock return predictions via quantum neural network and tensor network

Nozomu Kobayashi, Yoshiyuki Suimon, Koichi Miyamoto, Kosuke Mitarai

TL;DR

Both the quantum neural network and tensor network models have superior performances in the latest market environment, which suggests capability of model’s capturing non-linearity between input features.

Abstract

In this paper, we investigate the application of quantum and quantum-inspired machine learning algorithms to stock return predictions. Specifically, we evaluate the performance of quantum neural network, an algorithm suited for noisy intermediate-scale quantum computers, and tensor network, a quantum-inspired machine learning algorithm, against classical models such as linear regression and neural networks. To evaluate their abilities, we construct portfolios based on their predictions and measure investment performances. The empirical study on the Japanese stock market shows the tensor network model achieves superior performance compared to classical benchmark models, including linear and neural network models. Though the quantum neural network model attains a lowered risk-adjusted excess return than the classical neural network models over the whole period, both the quantum neural network and tensor network models have superior performances in the latest market environment, which suggests the capability of the model's capturing non-linearity between input features.

The cross-sectional stock return predictions via quantum neural network and tensor network

TL;DR

Both the quantum neural network and tensor network models have superior performances in the latest market environment, which suggests capability of model’s capturing non-linearity between input features.

Abstract

In this paper, we investigate the application of quantum and quantum-inspired machine learning algorithms to stock return predictions. Specifically, we evaluate the performance of quantum neural network, an algorithm suited for noisy intermediate-scale quantum computers, and tensor network, a quantum-inspired machine learning algorithm, against classical models such as linear regression and neural networks. To evaluate their abilities, we construct portfolios based on their predictions and measure investment performances. The empirical study on the Japanese stock market shows the tensor network model achieves superior performance compared to classical benchmark models, including linear and neural network models. Though the quantum neural network model attains a lowered risk-adjusted excess return than the classical neural network models over the whole period, both the quantum neural network and tensor network models have superior performances in the latest market environment, which suggests the capability of the model's capturing non-linearity between input features.
Paper Structure (14 sections, 25 equations, 7 figures, 3 tables)

This paper contains 14 sections, 25 equations, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Problem Definition
  4. Neural network
  5. Quantum circuit learning
  6. Tensor Network
  7. Optimization procedure
  8. Experiment
  9. Dataset
  10. Investment strategy
  11. Model architectures
  12. Backtesting result
  13. Conclusion and discussion
  14. Backtesting result from 2016

Figures (7)

  • Figure 1: The general structure of quantum circuit learning, where we have two quantum circuit architectures: Data encoding circuit $V(x)$ and parameterized quantum circuit $U(\theta)$
  • Figure 2: Our choice of a parameterized quantum circuit in the quantum circuit learning algorithm
  • Figure 3: The concept of our backtesting experiment, showing that we take three years as a training period and subsequent one year as a test period, rolling this process until the end of the backtesting period
  • Figure 4: The cumulative returns of portfolios constructed by various methods and that of TOPIX500
  • Figure 5: The cumulative excess returns of portfolios constructed by various methods over TOPIX500
  • ...and 2 more figures