Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe

AKSOE: Physik sozio-ökonomischer Systeme

AKSOE 10: Poster Session (posters are expected to be displayed the full day 8:30-18:00)

AKSOE 10.5: Poster

Mittwoch, 29. März 2006, 16:00–18:00, P2

To a local approximation method of time-series short-range predictability of developing assets — •Michael Romanovsky — Vavilov str., 38 119991 Moscow Russia

Two facts permit us to consider assets of developing stock markets as complex dynamical systems (or, at least, No-dqmore dynamicalNo-dq than the respect developed assets): a comparably small quantity of stocks and players and an influence of developed stock markets [1]. The method of short-range forecast is proposed for developing stock markets based on the generalized linear approximation of a day return. The method deals with the daily-recorded time series return of the investigated asset. The first step is the searching of predecessor time series parts, which have similar past values in comparison with the present part of the investigated time series. The short-range forecast (for one day and several days) is produced as some mean of these time series parts continued up to the forecast horizon. Thus we increase a redundant information of this time series [2]. The method gives satisfactory good several-days forecast: prediction of investigated time-series return changes (up or down) is larger than 60%, and the ratio of prediction error variation to the time-series variation is normally less than one. 1.T.B.Ersov, M.Yu.Romanovsky. Modern problems of statistical physics, 2, 168-179 (2003). 2.R.N.Mantegna, H.E.Stanley. An introductiion to econophysics. Correlation and complexity in finance. Cambrige University Press (Cambrige 2000).