A Dynamic Analysis of the Microstructure of Moving Average Rules in a Double Auction Market

Abstract

Inspired by the theoretically oriented dynamic analysis of moving average rules in the model of Chiarella, He, and Hommes (CHH) [Journal of Economic Dynamics and Control 30 (2006), 1729-1753, this paper conducts a dynamic analysis of a more realistic microstructure model of continuous double auctions in which the probability of heterogeneous agents trading is determined by the rules of either fundamentalists mean-reverting to the fundamental or chartists choosing moving average rules based on their relative performance. With such a realistic market microstructure, the model is able not only to obtain the results of the CHH model but also to characterize most of the stylized facts including volatility clustering, insignificant autocorrelations (ACs) of returns, and significant slowly decaying ACs of the absolute returns. The results seem to suggest that a comprehensive explanation of several statistical properties of returns is possible in a framework where both behavioral traits and realistic microstructure have a role.

Shock Transfer in Futures and Spot Markets: An Agent-Based Simulation Modelling Method

“…eir work, however, is more focused on how fundamental weight and chartist weight impact on the shock transfer progress, and they assume that every local agent uses a strategy that blends three components (fundamentalist, chartist, and noisy). Instead, we set the traders in our model to be fully fundamentalist or chartist as Chiarella et al [47] and change the experiment environments by changing the number of different types of agents, which is more intuitive and realistic. We have proved that our model can reproduce some important stylized facts in the real financial market.…”

“…Inspired by Ohi et al [37] and Torii et al [38], we built an order-driven cross-market model based on the same framework as Chiarella et al [47]. e salient features of the model of Chiarella et al [47] can be briefly outlined as follows: heterogeneous agents trade based on a fundamentalist or chartist strategy or noise trading strategy occasionally in the continuous double auction stock market where only one stock is traded. With a realistic market microstructure, their model can reproduce plenty of stylized facts in stock market, including the fat-tailed distribution of the stock's returns and the volatility clustering.…”

“…With a realistic market microstructure, their model can reproduce plenty of stylized facts in stock market, including the fat-tailed distribution of the stock's returns and the volatility clustering. In order to analyze how shocks to a single stock spread throughout the entire stock market by futures market and figure out the role of stock index futures during the crash periods, we extended the single-asset single-market model of Chiarella et al [47] to a multiasset multimarket model and introduced crossmarket traders into our artificial spot-futures markets. Most of the model structure is retained from the model of Chiarella et al [47] in terms of fundamentalists, chartists, and noise traders' trading strategies.…”

“…In order to analyze how shocks to a single stock spread throughout the entire stock market by futures market and figure out the role of stock index futures during the crash periods, we extended the single-asset single-market model of Chiarella et al [47] to a multiasset multimarket model and introduced crossmarket traders into our artificial spot-futures markets. Most of the model structure is retained from the model of Chiarella et al [47] in terms of fundamentalists, chartists, and noise traders' trading strategies. e detailed structure of our model is shown in Figure 1.…”

A behavioural model of investor sentiment in limit order markets

“…Different from the zero-intelligence models, the heterogenous agent models consider agents' strategic behaviors as potential explanations to the stylized facts. For example, chartist-fundamentalist models find that the chartist behavior contributes to fat tails, volatility clustering (see Anufriev and Panchenko (2009), Chiarella, Iori and Perellò (2009) and Chiarella, He and Pellizzari (2012)) and the diagonal effect in correlations of some order submission types (Kovaleva and Iori (2014)). However, replicating most of these stylized facts simultaneously remains very challenging for agent-based models.…”