J. Kwapien,S. Drozdz,J. Speth
Abstract:Based on the tick-by-tick stock prices from the German and American stock markets, we study the statistical properties of the distribution of the individual stocks and the index returns in highly collective and noisy intervals of trading, separately. We show that periods characterized by the strong inter-stock couplings can be associated with the distributions of index fluctuations which reveal more pronounced tails than in the case of weaker couplings in the market. During periods of strong correlations in the German market these distributions can even reveal an apparent Lévy-stable component.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: **quantitatively describe the possible relationship between stock price changes and the distribution characteristics of the corresponding index fluctuations**, especially the differences under the intensity of collective behavior in different markets. Specifically, by analyzing the high - frequency trading data of the German and US stock markets, the authors studied the statistical characteristics of individual stocks and index returns in highly collective and noisy trading intervals. They focused on the differences in the index fluctuation distribution characteristics between the periods of strong inter - stock coupling (i.e., when the market is highly volatile) and the periods of weak coupling (when the market is relatively calm), and how these differences are reflected in the stability at different time scales.
### Main research contents:
1. **Background introduction**: Based on a large number of historical datasets of stock prices and index values, previous studies have shown that the distributions of stock price fluctuations and index returns exhibit power - law scaling characteristics, and both have a similar scaling index \(\alpha\approx3.0\). However, according to the central limit theorem, if the data meet the conditions of independent and identically distributed and finite variance, its distribution should approach a normal distribution. Therefore, financial data violates the assumptions of the central limit theorem, which may be due to the correlations between the data.
2. **Methodology**: The authors used high - frequency trading data from 1998 - 1999, covering 30 companies in the Dow Jones Industrial Average (DJIA) and the German DAX30 index. By calculating the correlation matrices and their largest eigenvalues \(\lambda_1\) in different time windows, the authors divided the trading intervals into strongly correlated and weakly correlated categories and compared the distribution characteristics of index and individual stock returns in these two types of intervals.
3. **Result analysis**:
- **Distribution of index and individual stock returns**: On a short - time scale (e.g., 1 minute), the index return distribution in the DAX market shows obvious heavy - tailed characteristics, while the distribution in the DJIA market is relatively close to a normal distribution. As the time scale increases (e.g., 30 minutes), this difference gradually decreases.
- **Differences between strongly correlated and weakly correlated intervals**: In strongly correlated intervals, the distribution of index returns is more likely to be non - Gaussian, showing more obvious heavy - tailed characteristics; while in weakly correlated intervals, the distribution of index returns is closer to a Gaussian distribution.
- **Lévy stable distribution**: In some strongly correlated intervals, the index return distribution in the DAX market even exhibits the characteristics of Lévy stable distribution, indicating that there is strong collective behavior in the market in these intervals.
### Conclusion:
This study reveals the influence of the intensity of market collective behavior on the distribution characteristics of index fluctuations. Market fluctuations in strongly correlated intervals are more likely to exhibit non - Gaussian distributions, while fluctuations in weakly correlated intervals are closer to Gaussian distributions. This finding helps to better understand the dynamic behavior of financial markets, especially the fluctuation characteristics in different market environments.