Statistical properties of daily ensemble variables in the Chinese stock markets

Gao-Feng Gu,Wei-Xing Zhou

DOI: https://doi.org/10.1016/j.physa.2007.05.007

2006-03-18

Abstract:We study dynamical behavior of the Chinese stock markets by investigating the statistical properties of daily ensemble returns and varieties defined respectively as the mean and the standard deviation of the ensemble daily price returns of a portfolio of stocks traded in China's stock markets on a given day. The distribution of the daily ensemble returns has an exponential form in the center and power-law tails, while the variety distribution is log-Gaussian in the bulk followed by a power-law tail for large varieties. Based on detrended fluctuation analysis, R/S analysis and modified R/S analysis, we find evidence of long memory in the ensemble returns and strong evidence of long memory in the evolution of variety.

Physics and Society,Statistical Finance

What problem does this paper attempt to address?

This paper aims to study the dynamic behavior of the Chinese stock market and solve the following problems by analyzing the statistical characteristics of daily ensemble variables:

1. **Understanding the overall market behavior**: By studying the daily ensemble return and variety, the overall behavioral characteristics of the Chinese stock market as a complex system are revealed. The ensemble return is the mean of the price returns of a stock portfolio on a certain day, and the variety is the standard deviation of these returns.

2. **Exploring distribution characteristics**: Discuss the probability distribution characteristics of the ensemble return and variety. Specifically:
- The central part of the ensemble return distribution is exponential, and the tail is power - law distributed.
- The main part of the variety distribution is log - normal, and the large - value part shows a power - law tail.

3. **Long - term memory**: Detect whether there is long - term memory in the time series of the ensemble return and variety through detrended fluctuation analysis (DFA), R/S analysis, and improved R/S analysis. The research shows that:
- The ensemble return shows long - term memory in some time periods, but it is not significant in the entire time series.
- The variety shows strong long - term memory, and the Hurst index is approximately 0.91 to 0.93.

4. **Market efficiency controversy**: Through the above analysis, explore the efficiency problem of the Chinese stock market. The research shows that the Chinese stock market does not follow the random walk hypothesis in most time periods, especially showing strong long - term correlation in terms of variety.

### Formula summary

- **Daily log - return**:
\[
r_i(t)=\ln\left[\frac{P_i(t)}{P_i(t - 1)}\right]
\]

- **Ensemble return**:
\[
\mu(t)=\frac{1}{n}\sum_{i = 1}^n r_i(t)
\]

- **Variety**:
\[
\sigma^2(t)=\frac{1}{n}\sum_{i = 1}^n[r_i(t)-\mu(t)]^2
\]

- **Probability density function of ensemble return**:
\[
f(\mu)\sim\exp[-k_{\pm}\mu]
\]
where \(k_- = 76.1\pm3.3\) when \(- 0.06 <\mu<0\), \(k_+ = 83.2\pm3.5\) when \(0 <\mu<0.06\).

- **Power - law form of tail distribution**:
\[
f(\mu)\sim\mu^{-(1 +\alpha)}
\]
for the negative tail \(\alpha_- = 3.33\pm0.06\), for the positive tail \(\alpha_+ = 2.86\pm0.07\).

- **Probability density function of variety**:
\[
f(\sigma)\sim
\begin{cases}
\exp\left[-\frac{(\ln\sigma-\ln\sigma_0)^2}{2\text{Var}(\sigma)}\right]&\text{for not large }\sigma\\
\sigma^{-\beta}&\text{for large }\sigma
\end{cases}
\]
where \(\beta = 5.3\pm0.2\).

- **Fluctuation function of detrended fluctuation analysis**:
\[
F(s)\sim s^H
\]

Through these analyses, the paper attempts to reveal the complex dynamic behavior of the Chinese stock market and its similarities and differences with mature markets, thereby providing a theoretical basis for further understanding the characteristics and behaviors of emerging markets.