V. Alfi,F. Coccetti,M. Marotta,A. Petri,L. Pietronero
Abstract:We consider the role of finite size effects on the value of the effective Hurst exponent H. This problem is motivated by the properties of the high frequency daily stock-prices. For a finite size random walk we derive some exact results based on Spitzer's identity. The conclusion is that finite size effects strongly enhance the value of H and the convergency to the asymptotic value (H=1/2) is rather slow. This result has a series of conceptual and practical implication which we discuss.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the influence of the finite - size effect on the effective Hurst exponent \(H\), especially its manifestation in high - frequency daily stock prices. Specifically, the research motivation comes from the following aspects:
1. **Characteristics of high - frequency daily stock prices**: In high - frequency stock price data, intraday trading is relatively homogeneous, but large fluctuations at night (usually equivalent to the entire intraday fluctuation range) lead to an inherent limitation on the size of the data set.
2. **Influence of the finite - size effect**: These finite - size effects significantly change the determination of the effective Hurst exponent \(H\). In particular, when the data set is small, the value of \(H\) will be significantly higher than its asymptotic value \(H = 1/2\), and the speed of convergence to this asymptotic value is very slow.
3. **Explanation of deviations in practical applications**: Normally, values deviating from \(H = 1/2\) are considered to be caused by long - term correlations. However, this paper points out that such deviations may also be caused by the finite - size effect, especially in high - frequency financial data.
4. **Influence of non - stationarity**: Economic data usually shows obvious non - stationarity, which further slows down the speed at which \(H\) converges to \(H = 1/2\). Therefore, even long - time - series data may not reach the asymptotic value due to non - stationarity.
### Specific problems and methods
To study these problems, the author uses precise mathematical derivations, especially based on Spitzer's identity to calculate the expected value of the maximum of a random walk, and thereby derives the effective Hurst exponent \(H\) of a finite - size random walk. The main results include:
- **Finite - size effect enhances \(H\)**: The finite - size effect significantly enhances the value of \(H\), making it much higher than \(H = 1/2\), and the convergence speed is very slow.
- **Influence of different step - length distributions**: For step - length distributions with "fat - tailed" distributions, the finite - size effect is more significant because a few large step - lengths have a significant impact on the overall fluctuation.
- **Influence of non - stationarity**: Considering the non - stationarity of economic data, the speed at which \(H\) converges to \(H = 1/2\) may be even slower, thus providing another explanation for the \(H\) deviation observed in long - time - series data.
### Conclusion
This research reveals the important influence of the finite - size effect on the effective Hurst exponent, especially in high - frequency financial data. This not only helps to understand the deviation of the \(H\) value, but also provides a new perspective for analyzing financial time - series. Future research can consider incorporating these effects into more complex models for testing.