Direct evidence for inversion formula in multifractal financial
volatility measure
∗
Zhi-Qiang Jiang
1, 2
and Wei-Xing Zhou
1, 2, 3, 4
Email: wxzhou@ecust.edu.cn
1. School of Business, East China University of Science and Technology, Shanghai 200237, China
2. School of Science, East China University of Science and Technology, Shanghai 200237, China
3. Center for Econophysics Studies, East China University of Science and Technology, Shanghai 200237,
China
4. Research Center of Systems Engineering, East China University of Science and Technology, Shanghai
200237, China
Abstract
The inversion formula for conservative multifractal measures was unveiled mathematically a decade
ago, which is however not well tested in real complex systems. In this Letter, we propose to verify the
inversion formula using high-frequency turbulent financial data. We construct conservative volatility mea-
sure based on minutely S&P 500 index from 1982 to 1999 and its inverse measure of exit time. Both
the direct and inverse measures exhibit nice multifractal nature, whose scaling ranges are not irrelevant.
Empirical investigation shows that the inversion formula holds in financial markets.
Keywords: Econophysics, Multifractal analysis, Partition function approach, Inversion formula, Stock
markets
In recent years, the concept of inverse statistics has attracted much attention in turbulence [1, 2] as well
as in financial markets [3] based on time series analysis. The direct structure function concerns with the
statistical moments of a physical quantity g measured over a distance s such that S
q
(s) = hg
k
(s)
q
i. The
multifractal nature of direct structure functions has been well documented in turbulence [4, 5, 6], as well
as in finance [7, 8, 9], which is characterized by S
q
(s) ∼ s
ζ(q )
with a nonlinear scaling function ζ(q). In
contract, the inverse structure function is related to the exit distance, where the physical quantity fluctuation
exceeds a prescribed value, such that T
p
(g) = hs
p
(g)i. One can intuitively expect that there is a power law
scaling stating that T
p
(g) ∼ g
φ(p)
, where φ(p) is also a nonlinear function. Furthermore, if s ∼ g
1/h
, Schmitt
has shown that there is an inversion formula between the two types of scaling exponents such that ζ(q) = −p
and φ(p) = −q [10]. A similar intuitive derivation for the inversion formula is given for Laplacian random
walks [11].
The power-law scaling in inverse structure function was observed in the signals of two dimensional
turbulence [12, 13], in the synthetic velocity data of the GOY shell model [1, 14], and in the temperature and
longitudinal and transverse velocity data in grid-generated turbulence [15]. However, this scaling behavior
was not observed in other three dimensional turbulent flows from different experiments [2, 16, 17]. The
inversion formula for direct and inverse structure functions is verified for synthetic turbulence data of shell
models [14] but not for wind-tunnel turbulence data, which cover a range of Reynolds numbers Re = 400 −
1000 [16].
∗
Supported by the National Natural Science Foundation of China (Grant No. 70501011), the Fok Ying Tong Education Foundation
(Grant No. 101086), the Shanghai Rising-Star Program (Grant No. 06QA14015), and the Program for New Century Excellent Talents
in University (Grant No. NCET-07-0288).
1
http://www.paper.edu.cn