臺灣博碩士論文加值系統

Many technical and statistical tools have been proposed to predict stock market prices, including fundamental analysis (sns, 2002), technical analysis, traditional time series forecasting (Shumway
and Stoffer, 2000). Some statistical models (Kedem and Fokianos, 2002) consider the daily prices as discrete response variables; others consider treated daily prices as high-frequency financial time
series data, which are observations taken at fine time interval, and are summarized in Ghysels (2000) and Tsay (2002). However, few of these consistently achieve good result.
The characteristics of daily stock price data includes: discreteness with daily closing price, nonsuccessive with the interruption of holidays, not as high-frequency data as 5-minute transactions data, not as low-frequency data without interruption as weekly, monthly, or yearly stock price data. The purpose of this study is to develop a dynamical, semi-parametric regression (Cox, 1972) based on the multi-state model to analyze the daily stock price data. This model considers characteristics of the daily price, including the discreteness, three distinct states arising from the very short duration of successive “going up”, “falling down” or “remaining at the same”. Individuals are allowed to transform bidirectionally from one state to another state. Six transient (from one state to another) probabilities, and three survival (remaining at the same state) probabilities can be defined and estimated from the proposed model. Finally, this proposed model dynamically uses today’s financial information as covariates to predict the trend (state) of the daily stock price for the next day.
The total proportion of correct prediction will be interested and is used to evaluate the fitness of the model to a particular stock. For empirical evaluation of the model, 500 listed companies will be random selected. For each listed company, a prediction model including the same covariates will be fitted, Some the results are promising, the best proportion of the correct prediction of the trend of the daily stock price is 0.5628, the average overall correct prediction proportion is 0.4498 based on the randomly selected 500 list companies and two simple covariates: daily price and daily volume. The proposed model can be analyzed easily, with some modifications of the observed data, by the basic survival analysis in most standardized statistical softwares.

1 Introduction 1
2 Literature Review 6
2.1 Survival Analysis . . . . . . . . . . . . . . . . . . . 6
2.2 Multi-State Model . . . . . . . . . . . . . . . . . . . 13
2.3 Competing Risks Model . . . . . . . . . . . . . . . . . 13
2.4 The Illness-Death Model . . . . . . . . . . . . . . . . 14
3 Methods 17
3.1 Daily Stock Price Data and Multi-State Process .. . . . 17
3.2 A Simple Three-State Model . . . . . . . . .. . . . . . 21
3.3 Multi-State Model for Daily Stock Price . . . . . . . . 23
3.4 Semiparametric Model: Estimation and Prediction . . . . 28
4 Examples 30
4.1 United Microelectronics Corporation (UMC) Daily Stock Price Data from
2000-2005 . . . . . . . . . . . . . . . 30
4.2 Analysis of 500 listed companies in Taiwan Stock Exchange . . . . . . . . . . . . . . . . . . . . . . . . . 32
5 Discussion 37
References 39
Appendix 43
A 500 Listed Companies 43

[1] A. Dewanji, J. K. Kalbfleisch (1986), Nonparametric Methods for Survival/Sacrifice
Experiment, Biometrics, 42:325-341.
[2] A. E. Raftery (1985). A model for higher-order Markov chains. Journ of the Royal
Statistical Society. Series B, 47:528-539.
[3] B. Kedem and K. Fokianos (2002). Regression Models for Time Series Analysis,
Wiley, New York.
[4] C. S.Wong andW. K. Li (2000). On a mixture autoregressive model. Journal of the
Royal Statistical Society, Series B, 62:95-115
[5] C. S. Wong and W. K. Li (2001a). On a mixture logistic mixture autoregressive
model. Biometrika, 88:833-846.
[6] C. S. Wong and W. K. Li (2001b). On a mixture autoregressive conditional model
heteroscedastic model. Journal of the American Statistical Association, 96:982-995.
[7] D. Commenges (2002), Inference for multi-state models from interval-censored
data, Statistical Methods in Medical Research, 11:167-182
[8] D.R. Cox, (1972). Regression models and life tables, Journal of the Royal Statistical
Society, Series B, 34:187-202.
[9] E. Ghysels (2000). Some econometric Recipes for high-frequency data cocking.
Journal of Business and Economic Statistics, 18:154-163.
[10] E. L. Kaplan and P. Meier (1958). Nonparametric estimation from incomplete
observations. Journal of American Statistical Association, 3:457-481.
[11] E. V. Slud and B. Kedem (1994). Partial likelihood analysis of logistic regression
and autoregression. Statistica Sinica, 4:89-106.
[12] E .W . Saad, D .V . Prokhorov, and D .C . Wunsch(1998). Comparative Study
of Stock Trend Prediction Using Time Delay, Recurrent and Probabilistic Neural
Networks, IEEE TRANSACTIONS ON NEURAL NETWORKS, 9:1456-1470.
[13] G .S .Jang, F . Lai, B . W . Jiang,and L . H. Chien(1991), An Intelligent Trend Prediction
and Reversal Recognition System Using Dual-module Neural Networks,
IEEE,42-51.
[14] J. D. Kalbfleisch and R. L. Prentice (1973). Marginal likelihoods based on Cox’s
regression and life model, Biometrika, 60:267-278.
[15] J. Hauseman, A. Lo and C. MacKinlay (1992). An ordered probit analysis of transaction
stock prices, Journal of Financial Economics, 31:319-379.
[16] J .H . Wang ,and J .Y . Leu(1996). Stock Market Trend Prediction Using ARIMAbased
Neural Networks, IEEE, 4,2160-2165.
[17] J. R. Stroud, P. M¨ uller, and B. Sans´o (2001). Dynamic models for spatiotemporal
data. Journal of the Royal Statistical Society, Series B, 63:673-689.
[18] M.West, P. J. Harrison, and H. S. Migon (1985). Dynamic generalized linear models
and Bayesian forecasting. Journal of the American Statistical Association, 80:73-97,
with discussion.
[19] N. D. Le, R. D. Martin, and A. E. Reftery (1996). Modeling flat stretches, bursts,
and outliers in time series using mixture transition distribution models. Journal of
the American Statistical Association, 91:1504-1515.
[20] N. E. Breslow (1974). Covariance analysis of censored survival data, Biometrics,
30:89-99.
[21] O.O. Aalen (1976). nonparametric inference in connection with multiple decrements
models. Scandinavian Journal of Statistics, 3:15-27.
[22] P.K. Andersen and N. Keiding(2002), Multi-state models for event history analysis,
Statistical Methods In Medical Research, 11:91-115.
[23] P. K. Andersen, S. Z.Abildstrom, S.Rosth (2002), Competing risks as a multi-state
model, Statistical Methods in Medical Research, 11:203-215.
[24] R. E. McCulloch and R. S. Tsay (2001). Nonlinearity in high frequency data and
hierarchical models, Studies in Nonlinear Dynamics and Econometrics, 5:1-17.
[25] R. F. Engle and J. R. Russell (1998). Autoregressive conditional duration: A new
model of irregularly spaced transaction data, Econometrica, 66:1127-1162.
[26] R. H. Shumway and D. S. Stoffer (2000). Time Series Analysis and Its Applications.
Springer, New York.
[27] R. S. Tsay (2002). Analysis of Financial Time Series, Chapter 6,Wiley, New York.
[28] T. H. Rydberg and N. Shephard (1998). Dynamics of trade-by-trade price movements:
decomposition and models, Working paper, Nuffield College, Oxford
University.