DiffsFormer: A Diffusion Transformer on Stock Factor Augmentation

Paper: arXiv 2402.06656

Abstract

Machine learning models have demonstrated remarkable efficacy and efficiency in a wide range of stock forecasting tasks. However, the inherent challenges of data scarcity, including low signal-to-noise ratio (SNR) and data homogeneity, pose significant obstacles to accurate forecasting. To address this issue, we propose a novel approach that utilizes artificial intelligence-generated samples (AIGS) to enhance the training procedures. In our work, we introduce the Diffusion Model to generate stock factors with Transformer architecture (DiffsFormer). DiffsFormer is initially trained on a large-scale source domain, incorporating conditional guidance so as to capture global joint distribution. When presented with a specific downstream task, we employ DiffsFormer to augment the training procedure by editing existing samples. This editing step allows us to control the strength of the editing process, determining the extent to which the generated data deviates from the target domain. To evaluate the effectiveness of DiffsFormer augmented training, we conduct experiments on the CSI300 and CSI800 datasets, employing eight commonly used machine learning models. The proposed method achieves relative improvements of 7.2% and 27.8% in annualized return ratio for the respective datasets. Furthermore, we perform extensive experiments to gain insights into the functionality of DiffsFormer and its constituent components, elucidating how they address the challenges of data scarcity and enhance the overall model performance. Our research demonstrates the efficacy of leveraging AIGS and the DiffsFormer architecture to mitigate data scarcity in stock forecasting tasks.

Complexity vs Empirical Score

• Math Complexity: 8.0/10
• Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced mathematical concepts from diffusion models and transformer architectures, involving concepts like Markov chains, noise prediction, and conditional guidance, alongside detailed algorithmic descriptions. It demonstrates high empirical rigor with extensive experiments on real-world datasets (CSI300/CSI800), quantitative performance metrics (7.2% and 27.8% improvements in annualized return), and ablation studies on model components.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Address Data Scarcity in Stock Forecasting<br>Low SNR & Data Homogeneity"] --> B{"Key Methodology<br>Diffusion Model on Stock Factors<br>DiffsFormer Architecture"}
  
  B --> C["Inputs: Source Domain<br>CSI300 & CSI800 Datasets"]
  C --> D["Training Phase<br>Conditional Guidance<br>Captures Global Joint Distribution"]
  D --> E["Augmentation Phase<br>Conditional Editing of Samples<br>Controls Deviation Strength"]
  E --> F["Computational Process<br>8 ML Models + AIGS<br>Augmented Training"]
  
  F --> G["Outcomes & Findings<br>CSI300: +7.2% Annualized Return<br>CSI800: +27.8% Annualized Return<br>Data Scarcity Mitigated"]

Cited References (54)

1. Challenges of Developing AI Applications in the Evolving Digital World and Recommendations to Mitigate Such Challenges: A Conceptual View (2021)
2. Stock Market Prediction via Deep Learning Techniques: A Survey — → 2212.12717
3. The efficient market hypothesis and its critics (2003)
4. MTMD: Multi-Scale Temporal Memory Learning and Efficient Debiasing Framework for Stock Trend Forecasting (2022) — link
5. Anomalies across the globe: Once public, no longer existent? (2020)
6. Denoising diffusion probabilistic models (2020)
7. Glide: Towards photorealistic image generation and editing with text-guided diffusion models (2021) — link
8. Hierarchical text-conditional image generation with clip latents (2022) — link

1 of 54 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.