A1 Vertaisarvioitu alkuperäisartikkeli tieteellisessä lehdessä
Time-independent models of asset returns revisited
Tekijät: Gillemot L, Töyli J, Kertesz J, Kaski K
Kustantaja: Elsevier
Julkaisuvuosi: 2000
Journal: Physica A: Statistical Mechanics and its Applications
Tietokannassa oleva lehden nimi: PHYSICA A
Lehden akronyymi: PHYSICA A
Vuosikerta: 282
Numero: 1-2
Aloitussivu: 304
Lopetussivu: 324
Sivujen määrä: 21
ISSN: 0378-4371
eISSN: 1873-2119
DOI: https://doi.org/10.1016/S0378-4371(00)00101-1
Verkko-osoite: https://doi.org/10.1016/S0378-4371(00)00101-1
Tiivistelmä
In this study we investigate various well-known time-independent models of asset returns being simple normal distribution, Student t-distribution, Levy, truncated Levy, general stable distribution. mixed diffusion jump, and compound normal distribution. For this we use Standard and Poor's 500 index data of the New York Stock Exchange, Helsinki Stock Exchange index data describing a small volatile market, and artificial data. The results indicate that all models, excluding the simple normal distribution, are, at least. quite reasonable descriptions of the data. Furthermore. the use of differences instead of logarithmic returns tends to make the data looking visually more Levy-type distributed than it is. This phenomenon is especially evident in the artificial data that has been generated by an inflated random walk process. (C) 2000 Elsevier Science B.V. All rights reserved.
In this study we investigate various well-known time-independent models of asset returns being simple normal distribution, Student t-distribution, Levy, truncated Levy, general stable distribution. mixed diffusion jump, and compound normal distribution. For this we use Standard and Poor's 500 index data of the New York Stock Exchange, Helsinki Stock Exchange index data describing a small volatile market, and artificial data. The results indicate that all models, excluding the simple normal distribution, are, at least. quite reasonable descriptions of the data. Furthermore. the use of differences instead of logarithmic returns tends to make the data looking visually more Levy-type distributed than it is. This phenomenon is especially evident in the artificial data that has been generated by an inflated random walk process. (C) 2000 Elsevier Science B.V. All rights reserved.
Turun yliopiston Tutkimustietojärjestelmän portaali