Prompt Description

Create a diagram titled "Figure 1: Technical Architecture for Systemic Risk Early Warning and Stress Testing in the Stock Market." The diagram should employ a **bottom-up** layered structure, comprising the following modules: 1. **Bottom Layer: Data & Mechanism Foundation Layer** * **Left - Multi-Source Heterogeneous Data Processing**: Includes macro data (dimensionality increase/interpolation), high-frequency trading data (dimensionality reduction/aggregation), and sentiment text (LLM-based causal extraction). * **Right - Risk Evolution Mechanism Analysis**: Includes multi-relational spatio-temporal networks, RMT-based structure denoising, and Bayesian cross-scale causal inference. * **Output**: Construction of a "Risk Monitoring Indicator System" (external shocks + internal market factors). 2. **Middle Layer: Dual-Layer Monitoring & Warning Core** * **Left - Overall Market Early Warning (Macro-Level)**: * Multi-scale prediction with long and short cycles (LSTM+GARCH) * Causal-guided early warning (Attention-LSTM + PCMCI) * Deep state extraction (Recursive VAE) * **Right - Main Body Market Early Warning (Micro-Level)**: * Knowledge graph hidden association mining (GCN) * Spatio-temporal dynamic evolution (ST-DERU) * Extreme risk capture (GAN + Neural CDE + Patch linear embedding) 3. **Top Layer: Dynamic Stress Testing & Application Layer** * **Core Models**: EXformer (exogenous variable enhancement), PatchTST-ST, iTransformer-ST. * **Transmission Models**: GDP-STM (growth transmission), PPI/CPI-STM (inflation transmission). * **Final Output**: Dynamic stress scenario generation ➜ Financial Market Stress Index (FSI) prediction ➜ Risk level and decision support. 4. **Connecting Arrows**: * From the "Bottom Layer" upwards to the "Middle Layer," indicating data and theoretical support for model construction. * From the "Middle Layer" upwards to the "Top Layer," indicating that early warning signals are input into the stress testing model for simulation.