Activity autocorrelation in financial markets. A comparative study between several models

Luigi Palatella,Josep Perello,Miquel Montero,Jaume Masoliver

DOI: https://doi.org/10.1140/epjb/e2004-00161-6

2003-12-19

Abstract:We study the activity, i.e., the number of transactions per unit time, of financial markets. Using the diffusion entropy technique we show that the autocorrelation of the activity is caused by the presence of peaks whose time distances are distributed following an asymptotic power law which ultimately recovers the Poissonian behavior. We discuss these results in comparison with ARCH models, stochastic volatility models and multi-agent models showing that ARCH and stochastic volatility models better describe the observed experimental evidences.

Statistical Mechanics,Physics and Society,Statistical Finance

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the autocorrelation problem of financial market activities (i.e., the number of transactions occurring per unit time). Specifically, the authors studied the time series of trading activities in the financial market and used the Diffusion Entropy (DE) technique to analyze the autocorrelation in these time series.

### Main problems:
1. **Causes of activity autocorrelation**: The authors explored that the activity autocorrelation is caused by the time - distance distribution between peaks (following an asymptotic power law), and this distribution eventually returns to Poisson behavior.
2. **Model comparison**: The authors compared the experimental results with several different models, including the ARCH model, the stochastic volatility model, and the multi - agent model, to determine which model can better describe the observed experimental phenomena.

### Specific content:
- **Power - law distribution**: The authors found that the probability density function \(\psi(\tau)\) of the time interval \(\tau\) of trading activities usually presents an asymptotic power - law form:
\[
\psi(\tau) \sim \frac{1}{\tau^\beta}
\]
This indicates that trading activities have strong inhomogeneity.

- **Diffusion entropy analysis**: Through the diffusion entropy technique, the authors analyzed the time series of trading activities and found the abnormal scaling behavior of the activity distribution. Specifically, the expression of the diffusion entropy \(S(t)\) is:
\[
S(t)=-\int_{-\infty}^{\infty} p(y, t) \ln [p(y, t)] dy = A+\delta \ln(t)
\]
where \(\delta\) is a scaling parameter used to measure the memory of the time series.

- **Model verification**: The authors used the TARCH model and the stochastic volatility model to generate artificial time series and performed diffusion entropy analysis on these sequences. The results show that both models can well reproduce the scaling behavior in the actual data. In particular, the TARCH model shows a scaling index of \(\delta = 0.90\) on a short - time scale, which is consistent with the actual data.

### Conclusion:
The authors believe that the TARCH and stochastic volatility models can more accurately describe the activity autocorrelation in the financial market than the on - off intermittency model based on random walks. The main reason is that these models can capture the power - law characteristics of the time - interval distribution between trading activity peaks, while the on - off intermittency model is affected by the random - walk assumption, resulting in its predicted scaling behavior being inconsistent with the actual data.

In summary, this paper aims to reveal the activity autocorrelation in the financial market and the mechanisms behind it, and through the comparison of multiple models, has found an interpretation method that is more in line with the actual data.