Behavioral Prior Elicitation via LLM for Heterogeneous-Agent Bitcoin Flash-Crash Simulation
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
Related
- bayprior: Structured Bayesian Prior Elicitation, Conflict Diagnostics, and Regulatory Reporting
- Weighting risk impact indicators in a food system: expert elicitation and analysis prior to inclusion in a multi-risk assessment
- A Least-Squares Approach to Sample-Based Prior Elicitation
- Spectra: Exact Component Transport for Test-Time Prior Adaptation in Simulation-Based Inference
- Do Ray-Traced Cities Make a Better Channel Prior for an Unseen Site? A Pre-Registered Study in Simulation, and a Test on Measured Channels that Reverses It
BibTeX
@misc{pham2026behavioral,
title = {{Behavioral Prior Elicitation via LLM for Heterogeneous-Agent Bitcoin Flash-Crash Simulation}},
author = {Tran Phuong Linh Pham and Thi Minh Phuong Ngo and Phuong Thao Nguyen and Quang Minh Nguyen and Thai Hoa Ngo and Trong Nghia Nguyen},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.03660},
eprint = {2611.03660},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.03660},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Behavioral Prior Elicitation via LLM for Heterogeneous-Agent Bitcoin Flash-Crash Simulation
AU - Tran Phuong Linh Pham
AU - Thi Minh Phuong Ngo
AU - Phuong Thao Nguyen
AU - Quang Minh Nguyen
AU - Thai Hoa Ngo
AU - Trong Nghia Nguyen
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.03660
UR - https://arxiv.org/abs/2611.03660
AB - 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
ER -