Multi-period Learning for Financial Time Series Forecasting

TL;DR

The paper addresses the challenge of financial time series forecasting by leveraging multi-period inputs that capture short-, medium-, and long-term information. It proposes the Multi-period Learning Framework (MLF), which integrates three novel components—Inter-period Redundancy Filtering (IRF), Multi-period self-Adaptive Patching (MAP), and Learnable Weighted-average Integration (LWI)—along with a Patch Squeeze module to enhance efficiency. Through experiments on fund sales data and public TSF datasets, MLF consistently outperforms both single-period baselines and existing multi-period methods, while also delivering notable efficiency gains and successful deployment in a production setting. These results demonstrate the practicality of multi-period designs for accurate and efficient financial forecasting.

Abstract

Time series forecasting is important in finance domain. Financial time series (TS) patterns are influenced by both short-term public opinions and medium-/long-term policy and market trends. Hence, processing multi-period inputs becomes crucial for accurate financial time series forecasting (TSF). However, current TSF models either use only single-period input, or lack customized designs for addressing multi-period characteristics. In this paper, we propose a Multi-period Learning Framework (MLF) to enhance financial TSF performance. MLF considers both TSF's accuracy and efficiency requirements. Specifically, we design three new modules to better integrate the multi-period inputs for improving accuracy: (i) Inter-period Redundancy Filtering (IRF), that removes the information redundancy between periods for accurate self-attention modeling, (ii) Learnable Weighted-average Integration (LWI), that effectively integrates multi-period forecasts, (iii) Multi-period self-Adaptive Patching (MAP), that mitigates the bias towards certain periods by setting the same number of patches across all periods. Furthermore, we propose a Patch Squeeze module to reduce the number of patches in self-attention modeling for maximized efficiency. MLF incorporates multiple inputs with varying lengths (periods) to achieve better accuracy and reduces the costs of selecting input lengths during training. The codes and datasets are available at https://github.com/Meteor-Stars/MLF.

Figures (11)

• Figure 1: Current TSF scheme (a) and ours (b). The blue rectangle highlights the characteristics of the multi-period inputs.
• Figure 2: Comparison between multi-period inputs in subfigure (c) and multi-scale inputs in subfigure (b).
• Figure 3: MLF and corresponding components.
• Figure 4: Illustration of the standard multi-period patching (using fixed patch lengths) and self-adaptive multi-period patching (using self-adaptively varied patch lengths).
• Figure 5: Illustration of patch squeeze module.