Forecasting Methods And Stock Market Analysis (Page 6 of 8) in pdf

108

Virginica Rusu and Cristian Rusu

The identiﬁcation of ARMA models is based on both the autocorrelation

and partial autocorrelation functions, and the informational criteria, used

by some of the specialized forecasting software. Some of the informational

criteria are AIC (Akaike Information Criteria), BIC (Bayesian Information

Criteria), Schwarz and Φ.

One of the most frequently used and precise analysis and short-term fore-

cast techniques is known as the Box-Jenkins method, based on the con-

cept of ARIMA (Integrated Mixed Autoregressive – Moving Average Series)

process.

The ARMA(p,q) model is suitable for modeling stationary processes. A

stationary process features a process generation mechanism that is invariant

in time. The average and the variance of a stationary process do not change

in time, and the covariance of the variables depends only on the length of the

time interval that separates the two variables. The trend & seasonal compo-

nents do not occur in stationary series. The non-stationary ARIMA(p,d,q)

models are speciﬁc to the non-seasonal phenomena whose trend can be elim-

inated, and thus the process can be made stationary, by ﬁnite diﬀerences of

a certain order d. The stages of building an ARIMA model are: the identi-

ﬁcation of the model, the assessment of the parameters, the check-up of the

model selected, the utilization of the model for forecast.

The linear models are not always suﬃcient to satisfactorily model real

world phenomena. There is an increasing interest in developing, including

for modeling time series. Some of the interesting features that the non-

linear models may oﬀer are: (1) the prediction interval does not increase

in time, and (2) the distribution of forecast errors is not a normal one.

Several non-linear models are frequently used: statistical models of non-

linear structure (non-linear autoregressive processes, threshold models, bi-

linear models), models for changing variance, and neural networks.

Comparative studies by diﬀerent authors, between the neural network-

based methods, and other methods, showed that:

• automatic application of the neural network process, as a “black box ”,

leads to weak results,

• the neural networks can be applied with satisfactory results only on ex-

tended time series,

• for the time being, the eﬀort that the use of neural networks requires (eﬀort

in computation, method comprehension and result assessment) appears to

lead to insigniﬁcant improvements of the forecast results, at least for the

moment,

• the neural networks are unable to predict the dramatic changing of the

shares’ quotations; other factors of inﬂuence should be taken into consid-

eration (news about the economic and political situation).

Forecasting Methods And Stock Market Analysis Page 6