Paper: arXiv 2611.03660

Authors: Tran Phuong Linh Pham, Thi Minh Phuong Ngo, Phuong Thao Nguyen, Quang Minh Nguyen, Thai Hoa Ngo, Trong Nghia Nguyen

Abstract

Bitcoin perpetual futures trade continuously with shallow liquidity and high price impact, making them prone to millisecond-scale flash crashes. Agent-based limit order book simulators are a natural tool for studying such events, but their behavioral parameters are usually hand-crafted or require full order book histories that are rarely available at scale. We propose a four-stage pipeline that constructs a reproducible 100 ms BTCUSDT flash-crash event panel from 66 detected events, elicits behavioral priors for four agent archetypes using an offline large language model conditioned on empirical microstructure anchors, instantiates a heterogeneous limit order book simulator with the validated priors, and applies causal discovery plus counterfactual analysis to the resulting synthetic panel. Across 500 simulation runs, LLM-elicited priors produce a crash rate of 0.176 versus 0.008 under uninformative uniform priors and generate heav- ier return tails. In the frozen counterfactual analysis, reducing leverage amplification by one within-sample standard deviation lowers the predicted crash rate from 0.176 to 0.136, whereas reducing directional pressure yields only a marginal change. These find- ings suggest that empirically anchored LLM elicitation can improve simulator fidelity and help identify plausible crash-amplification mechanisms in a controlled synthetic environment

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents a novel methodology for behavioral prior elicitation using LLMs, anchored by robust empirical data and a well-structured simulation framework. The combination of advanced simulation techniques with causal discovery and counterfactual analysis demonstrates strong empirical rigor and a high degree of novelty.

Research Flowchart

  flowchart TD
    A[Research Goal: Simulate Bitcoin Flash Crashes & Identify Mechanisms] --> B{Methodology: Four-Stage Pipeline};
    B --> C[Stage 1: Construct 100ms Flash-Crash Event Panel];
    C --> D[Data Input: 66 Detected BTCUSDT Flash-Crash Events];
    D --> E[Stage 2: Elicit Behavioral Priors via LLM];
    E --> F[Computational Process: Offline LLM Conditioned on Empirical Microstructure];
    F --> G[Stage 3: Instantiate Heterogeneous LOB Simulator with Validated Priors];
    G --> H[Computational Process: 500 Simulation Runs];
    H --> I[Stage 4: Causal Discovery & Counterfactual Analysis];
    I --> J[Key Finding 1: LLM Priors Yield Crash Rate of 0.176 vs 0.008 (Uniform Priors)];
    J --> K[Key Finding 2: LLM Priors Generate Heavier Return Tails];
    K --> L[Key Finding 3: Reducing Leverage Amplification Lowers Crash Rate (0.176 to 0.136)];
    L --> M[Key Finding 4: Directional Pressure Reduction has Marginal Impact];