arXivDaily arXiv每日学术速递 周一至周五更新
arXiv physics/0112045physics.soc-phphysics.data-anq-fin.ST
收藏

市场指数的动力学、马尔可夫链和随机行走问题

Dynamics of market indices, Markov chains, and random walking problem

M. I. Krivoruchenko

摘要:

美国主要市场指数DJIA,S&P,纳斯达克和纽约证券交易所的动态分析的角度来看,随机行走问题的两步相关的市场走势。表征随机动态的参数根据每日、每周和每月系列的历史报价经验确定。结果表明,存在统计上显着的相关性,随后的市场走势。周参数和月参数是根据日参数计算的,假设具有两步相关性的马尔可夫链给出了市场随机动力学的完整描述。我们表明,宏观和微观参数服从雷诺群方程。通过将回归方程中的参数与历史值进行比较,发现马尔可夫链方法对周行情的预测是合理的,但低估了月行情继续下跌的可能性。计算了“买入并持有”和“跟随趋势”策略的收益率及其离散度。讨论了如何将这两种策略结合起来以减小色散的问题,并给出了解析解。报告了将系列分析与蜡烛图技术相结合的“基于计算机的”策略的构建结果。

英文摘要:

Dynamics of the major USA market indices DJIA, S&P, Nasdaq, and NYSE is analyzed from the point of view of the random walking problem with two-step correlations of the market moves. The parameters characterizing the stochastic dynamics are determined empirically from the historical quotes for the daily, weekly, and monthly series. The results show existence of statistically significant correlations between the subsequent market moves. The weekly and monthly parameters are calculated in terms of the daily parameters, assuming that the Markov chains with two-step correlations give a complete description of the market stochastic dynamics. We show that the macro- and micro-parameters obey the renorm group equation. The comparison of the parameters determined from the renorm group equation with the historical values shows that the Markov chains approach gives reasonable predictions for the weekly quotes and underestimates the probability for continuation of the down trend in the monthly quotes. The return and its dispersion for the "buy-and-hold" and "follow-trend" strategies are calculated. The problem of how to combine these two strategies to reduce the dispersion is discussed and its analytical solution is proposed. The results of the constructing a "computer-based" strategy which combines the series analysis with the candlesticks charting techniques are reported.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑