arXiv subjects
Publications and source records attributed to Hideki Takayasu.
We introduce an autoregressive-type model of prices in financial market taking into account the self-modulation effect. We find that traders are mainly using strategies with weighted feedbacks of past prices. These feedbacks are responsible for the slow diffusion in short times, apparent trends and power law distribution of price changes.
Following the work of Okuyama, Takayasu and Takayasu [Okuyama, Takayasu and Takayasu 1999] we analyze huge databases of Japanese companies' financial figures and confirm that the Zipf's law, a power law distribution with the exponent -1, has been maintained over 30 years in the income distribution of Japanese companies with very high precision. Similar power laws are found not only in income distribution of company's income, but also in the distributions of capital, sales and number of employees. From the data we find an important time evolutionary property that the growth rate of income is approximately independent of the value of income, namely, small companies and large ones have similar statistical chances of growth. This observational fact suggests the applicability of the theory of multiplicative stochastic processes developed in statistical physics. We introduce a discrete version of Langevin equation with additive and multiplicative noises as a simple time evolution model of company's income. We test the validity of the Takayasu-Sato-Takayasu condition [Takayasu, Sato and Takayasu 1997] for having an asymptotic power law distribution as a unique statistically steady solution. Directly estimated power law exponents and theoretically evaluated ones are compared resulting a reasonable fit by introducing a normalization to reduce the effect of gross economic change.
We introduce a mean-field type approximation for description of company's income statistics. Utilizing huge company data we show that a discrete version of Langevin equation with additive and multiplicative noises can appropriately describe the time evolution of a company's income fluctuation in statistical sense. The Zipf's law of income distribution is shown to be hold in a steady-sate widely, and country-dependence of income distribution can also be nicely implemented in our numerical simulation.
We first estimate the average growth of a company's annual income and its variance by using both real company data and a numerical model which we already introduced a couple of years ago. Investment strategies expecting for income growth is evaluated based on the numerical model. Our numerical simulation suggests the possibility that an investment strategy focusing on the medium-sized companies gives the best asset growth with relatively low risk.
For the purpose of elucidating the correlation among currencies, we analyze daily and high-resolution data of foreign exchange rates. There is strong correlation for pairs of currencies of geographically near countries. We show that there is a time delay of order less than a minute between two currency markets having a strong cross-correlation. The cross-correlation between exchange rates is lower in shorter time scale in any case. As a corollary we notice a kind of contradiction that the direct Yen-Dollar rate significantly differs from the indirect Yen-Dollar rate through Euro in short time scales. This result shows the existence of arbitrage opportunity among currency exchange markets.
We analyze high-resolution foreign exchange data consisting of 20 million data points of USD-JPY for 13 years to report firm statistical laws in distributions and correlations of exchange rate fluctuations. A conditional probability density analysis clearly shows the existence of trend-following movements at time scale of 8-ticks, about 1 minute.
We first show that there are in fact triangular arbitrage opportunities in the spot foreign exchange markets, analyzing the time dependence of the yen-dollar rate, the dollar-euro rate and the yen-euro rate. Next, we propose a model of foreign exchange rates with an interaction. The model includes effects of triangular arbitrage transactions as an interaction among three rates. The model explains the actual data of the multiple foreign exchange rates well.
We analyze tick data of yen-dollar exchange with a focus on its up and down movement. We show that there exists a rather particular conditional probability structure with such high frequency data. This result provides us with evidence to question one of the basic assumption of the traditional market theory, where such bias in high frequency price movements is regarded as not present. We also construct systematically a random walk model reflecting this probability structure.
Analyzing historical data of price indices we find an extraordinary growth phenomenon in several examples of hyper-inflation in which price changes are approximated nicely by double-exponential functions of time. In order to explain such behavior we introduce the general coarse-graining technique in physics, the Monte Carlo renormalization group method, to the price dynamics. Starting from a microscopic stochastic equation describing dealers' actions in open markets we obtain a macroscopic noiseless equation of price consistent with the observation. The effect of auto-catalytic shortening of characteristic time caused by mob psychology is shown to be responsible for the double-exponential behavior.
We constructed an analog electrical circuit which generates fluctuations in which probability density function has power law tails. In the circuit fluctuations with an arbitrary exponent of the power law can be obtained by adjusting the resistance. With this low cost circuit the random fluctuations which have the similar statistics to foreign exchang rates can be generated as fast as an expensive digital computer.
A model of fluctuations in the market price including many deterministic dealers, who predict their buying and selling prices from the latest price change, is developed. We show that price changes of the model is approximated by ARCH(1) process. We conclude that predictions of dealers affected by the past price changes cause the fat tails of probability density function. We believe that this study bridges stochastic processes in econometrics with multi-agent simulation approaches.
We performed a network traffic simulation to clarify the mechanism producing self-similar traffic originating in the transport layer level. Self-similar behavior could be observed without assuming a long-tailed distribution of the input file size. By repeating simulations with modified TCP we found that the feedback mechanism from the network, such as packet transmission driven by acknowledgement packets, plays an essential role in explaining the self-similarity observed in the actual traffic.
We perform a simplified Ethernet traffic simulation in order to clarify the physical mechanism of the phase transition behavior which has been experimentally observed in the flow density fluctuation of Internet traffic. In one phase traffics from nodes connected with an Ethernet cable are mixed, and in the other phase, the nodes alternately send bursts of packets. The competition of sending packets among nodes and the binary exponential back-off algorithm are revealed to play important roles in producing $1/f$ fluctuations at the critical point.
We constructed an analog circuit generating fluctuations of which a probability density function has power law tails. In the circuit fluctuations with an arbitrary exponent of the power law can be obtained by tuning the resistance. A theory of a differential equation with both multiplicative and additive noises which describes the circuit is introduced. The circuit is composed of a noise generator, an analog multiplier and an integral circuit. Sequential outputs of the circuit are observed and their probability density function and autocorrelation coefficients are shown. It is found that correlation time of the autocorrelation coefficient is dependent on the power law exponent.
We analyze the distribution of income and income tax of individuals in Japan for the fiscal year 1998. From the rank-size plots we find that the accumulated probability distribution of both data obey a power law with a Pareto exponent very close to -2. We also present an analysis of the distribution of the debts owed by bankrupt companies from 1997 to March, 2000, which is consistent with a power law behavior with a Pareto exponent equal to -1. This power law is the same as that of the income distribution of companies. Possible implications of these findings for model building are discussed.
We propose a simple stochastic model of cascading transport in wave number space to clarify the origin of intermittent behavior of fully-developed fluid turbulence. In spite of lack of nonlinearity and viscosity the model gives non-Gaussian fluctuations and multifractal scalings consistent with experimental data.
Distribution functions of relative velocities among particles in a vibrated bed of powder are studied both numerically and theoretically. In the solid phase where granular particles remain near their local stable states, the probability distribution is Gaussian. On the other hand, in the fluidized phase, where the particles can exchange their positions, the distribution clearly deviates from Gaussian. This is interpreted with two analogies; aggregation processes and soft-to-hard turbulence transition in thermal convection. The non-Gaussian distribution is well-approximated by the t-distribution which is derived theoretically by considering the effect of clustering by inelastic collisions in the former analogy.
A set of hard spheres with tangential inelastic collision is found to reproduce observations of real and numerical granular matter. After time is scaled so as to cancel energy dissipation due to inelastic collisions out, inelastically colliding hard spheres in two dimensional space come to have $1/f^α$ fluctuation of total energy, non-Gaussian distribution of displacement vectors, and convective motion of spheres, which hard spheres with elastic collision, a conventional model of granular matter, cannot reproduce.