RAGIC: Risk-Aware Generative Model for Interval Construction

🚀 RAGIC has been accepted by IEEE Transactions on Knowledge and Data Engineering (TKDE) 2025! 🎉

🔹 Overview

RAGIC (Risk-Aware Generative model for Interval Construction) is a novel deep learning framework designed for stock market interval prediction, addressing the stochastic nature of financial markets. Unlike traditional point prediction, RAGIC forecasts a price range (lower and upper bounds), providing uncertainty-aware and risk-sensitive predictions.

🌟 Key Features

• Risk-Aware Forecasting: Integrates the volatility index to dynamically adjust interval width based on market risk.
• Generative Model for Market Scenarios: Uses GANs to simulate diverse market conditions, including extreme events.
• Two-Phase Framework:
  1. Sequence Generation: GAN-based model learns historical stock patterns and simulates future price sequences.
  2. Interval Construction: A horizon-wise strategy gathers predictions across different time horizons, adjusting risk-sensitive intervals adaptively.
• Lightweight & Scalable: Trained once, reusable for multiple sequence generations, relying on public data only.
• State-of-the-Art Performance: Achieves 95%+ coverage with narrow, informative intervals, outperforming baselines on global stock indices.

---

📖 Methodology

1️⃣ Sequence Generation (GAN-based Forecasting)

RAGIC employs a GAN model to generate diverse future price sequences by learning historical market features. The generator consists of:

• Risk Module: Captures market risk via a risk attention score derived from the volatility index (e.g., VIX for SPX, VXD for DOW).
• Temporal Module: Extracts multi-scale trends from historical prices, capturing both local and global patterns.
• Latent Space Randomness: Introduces stochasticity throughout the model layers, ensuring realistic financial variability.

---

2️⃣ Interval Construction (Risk-Sensitive Prediction Intervals)

• Implements a horizon-wise strategy to aggregate multiple simulated sequences.
• Uses statistical inference to construct adaptive intervals with width based on market risk.
• A wider interval signals higher market volatility, aiding in risk assessment and hedging strategies.

---

🚀 Contributions

✅ First generative model for interval prediction in finance
✅ Risk-sensitive interval adjustment using volatility index insights
✅ Superior performance: Consistently achieves 95%+ coverage with balanced informativeness
✅ Scalable & practical: Trained once, adaptable across global stock indices

---

The code will be released soon.

Citation

If you use this work for your research, please kindly cite our paper:

@article{gu2025ragic,
  title={RAGIC: Risk-Aware Generative Framework for Stock Interval Construction},
  author={Gu, Jingyi and Du, Wenlu and Wang, Guiling},
  journal={IEEE Transactions on Knowledge and Data Engineering},
  year={2025},
  publisher={IEEE},
  doi={10.1109/TKDE.2025.3533492}
}