An empirical behavioural order-driven model with price limit rules

Gao-Feng Gu, Xiong Xiong, Hai-Chuan Xu, Wei Zhang, Yongjie Zhang, Wei Chen, Wei‐Xing ZhouarXiv 2017 · DOI · arXiv

Abstract

This paper presents an Empirical Behavioural Order-Driven (EBOD) model designed to simulate stock market dynamics under price limit constraints. By incorporating empirical regularities—such as long memory in order directions and relative prices, asymmetric price distributions, and nonlinear order size dependencies—the model successfully replicates real-world "stylized facts." Crucially, computational experiments demonstrate that asymmetric price limits (where upper and lower bounds differ) lead to catastrophic market outcomes, causing stock prices to either explode toward infinity or vanish entirely.

The Regulatory Challenge of Price Limits

In many emerging markets, regulators implement price limits to curb excessive volatility and speculation. However, the interaction between these hard boundaries and trader behavior is complex. Most traditional models fail to account for how traders adjust their order placement and cancellation strategies as prices approach these limits. This research addresses a specific regulatory concern: the impact of asymmetric price limits, such as the +1% up and -5% down rule proposed for "risk-warning" stocks in the Chinese market in 2012.

Building the EBOD Model

The EBOD model decomposes market activity into two primary empirical processes calibrated using high-frequency data from the Shenzhen Stock Exchange (SZSE):

  1. Order Placement: Unlike "zero-intelligence" models, the EBOD model uses empirical distributions to determine order characteristics. A central component is the relative price ($x$), which is redefined to integrate the physical boundaries of price limits. The model accounts for the asymmetric distribution of these prices and the nonlinear relationship between price aggressiveness and order size.
  2. Order Cancellation: This is modeled as a Poisson process where the probability of an order being cancelled depends on its relative price level and its position within the queue (temporal priority).
Figure 2: Empirical distribution of relative prices f(x) for four representative stocks, showing the asymmetric nature of order placement near price limits.

Empirical Foundation and Market Memory

The model is grounded in microscopic regularities observed in 2003 SZSE data (32 A-shares and 11 B-shares). A key finding is the presence of "long memory" in both order directions and relative prices. Using Detrended Moving Average (DMA) analysis, researchers calculated the Hurst exponent ($H$). A value of $H > 0.5$ indicates persistence, meaning past trends strongly influence future behavior.

Figure 1: (a) Fluctuation function F(l) vs. scale l for order directions, where the slope represents the Hurst exponent. (b) Histogram of Hurst exponents (Hs) for order directions across the dataset, showing values significantly above 0.5.

As shown in Figure 1, the Hurst exponents for order directions ($H_s$) typically fall between 0.75 and 0.9, indicating powerful herding behavior or persistence in investor sentiment. This memory is a critical driver of the emergent complexity seen in the simulated market.

The Destabilizing Effect of Asymmetry

The most significant result of the EBOD simulations is the mathematical proof of the danger of asymmetric price limits. When the upward limit is higher than the downward limit ($|\phi_+| > |\phi_-|$), the stock price diverges exponentially. Conversely, when the downward limit is larger (as in the 2012 proposal), the stock price rapidly decays toward zero.

Beyond price trajectories, the model also captures the "magnet effect," where the proximity to a price limit attracts more aggressive orders, potentially accelerating a crash or bubble rather than preventing it. Despite its reliance on simple micro-rules, the model naturally produces complex stylized facts like volatility clustering and heavy-tailed return distributions (following the "inverse cubic law").

Conclusion

The EBOD model provides a robust platform for policy experimentation. It demonstrates that price limits are not neutral tools; their specific configuration—especially symmetry—is vital for market stability. While the model currently treats the market as an aggregate of behaviors, future iterations could distinguish between institutional and retail participants to further refine its predictive power for modern, high-frequency trading environments.

Summary updated

Abstract

We develop an empirical behavioural order-driven (EBOD) model, which consists of an order placement process and an order cancellation process. Price limit rules are introduced in the definition of relative price. The order placement process is determined by several empirical regularities: the long memory in order directions, the long memory in relative prices, the asymmetric distribution of relative prices, and the nonlinear dependence of the average order size and its standard deviation on the relative price. Order cancellation follows a Poisson process with the arrival rate determined from real data and the cancelled order is determined according to the empirical distributions of relative price level and relative position at the same price level. All these ingredients of the model are derived based on the empirical microscopic regularities in the order flows of stocks on the Shenzhen Stock Exchange. The model is able to produce the main stylized facts in real markets. Computational experiments uncover that asymmetric setting of price limits will cause the stock price diverging exponentially when the up price limit is higher than the down price limit and vanishing vice versus. We also find that asymmetric price limits have influences on stylized facts. Our EBOD model provides a suitable computational experiment platform for academics, market participants and policy makers.
BibTeX
@misc{gu2017empirical,
  author = {Gao-Feng Gu and Xiong Xiong and Hai-Chuan Xu and Wei Zhang and Yongjie Zhang and Wei Chen and Wei‐Xing Zhou},
  title = {{An empirical behavioural order-driven model with price limit rules}},
  year = {2017},
  eprint = {1704.04354},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1704.04354},
}

Loading related content...

Figure