NECSI LOCAL SEMINAR SERIES PRESENTS:

"Statistical Physics Approaches to Financial Fluctuations"
By Fengzhong Wang

Abstract:
Financial fluctuation has long been a focus of economics and econo-physics research. To comprehensively study it, we investigate the intraday data for all stocks traded in the United States markets and focus on their volatilities. We explore the temporal structure in the series with a new approach, the return interval of volatility, which is the time interval between volatilities above a certain threshold. The distributions of return intervals are found to be a good scaling over a wide range of thresholds, time scales and markets. The scaling function is well-approximated to a stretched exponential. Further, we construct "clusters" of return intervals by their values and show that the distribution of cluster size has a much longer tail than that of the shuffled sequence, which indicates a strong memory effect in the original sequence. Using Detrended Fluctuation Analysis (DFA) method, we examine the correlations in the volatility and return interval series, and find that both types of series have similar strong long- term correlations. Moreover, we simulate the volatility with two wide-used models, FIGARCH and fractional Brownian motion (fBm). fBm captures the scaling in the distribution very well but FIGARCH does not. For the memory effect, both models are close to the empirical data. FIGARCH slightly overestimates and fBm slightly underestimates.

Speaker:
Mr. Wang is a Ph.D. candidate in Center for Polymer Studies and Physics Department at Boston University. His research focuses on the time series analysis and modeling of the financial markets. He also analyzes economic systems from the network approach. Mr. Wang has published more than 10 papers in peer journals, and served as the referee for 3 peer journals.

Attendence:
The Local Seminar Series is free and open to the public.