Wessa.net - Free Statistics and Forecasting Software (Calculators) v.1.2.1

All rights reserved. The non-commercial (academic) use of this software is free of charge. The only thing that is asked in return is to cite this software when results are used in publications.

Web-Enabled Scientific Services & Applications
(C)opyright 2002-2024
written by Prof. dr. P. Wessa

Server date: November 15, 2024, 4:46 am

This complicated online software application is the result of our continuous Research & Development efforts in the field of multidisciplinary Time Series Analysis in general and Financial Econometrics in particular. At the moment we cannot offer a complete reference manual due to the fact that the software is rapidly evolving, offering many new features and methods in a short period of time. A few simple examples however can be found here.

Below you can find an incomplete list of the features that are currently available in this module. Please, feel free to drop us a line if you have any suggestions or questions.

Example of Box-Jenkins Time Series Analysis
1	select Browser or Excel in the first selection box (this specifies how the output is displayed)
2	select Airline in the [data series] selection box
3	select Spectrum/(P)ACF in the [command] selection box
4	set L = 0 (this is the 'lambda' value of the Box-Cox transformation)
5	set d = 1 (this is the non-seasonal differencing order)
6	set D = 1 (this is the seasonal differencing order)
7	set s = 12 (default value; this is the number of periods per year = seasonality)
8	set K = 37 (this is the number of time lags to be computed)
9	click the lower Execute button
10	Now we can identify the ARMA model of the stationary time series. In this case we identify a non-seasonal MA(1) and a seasonal SMA(1) model. Hence, q=1 and Q=1.

General Commands
Edit data	Edit time series data (in textbox).
Meta data	Edit the meta data about the time series.

Specific Diagnostic Tools in Box-Jenkins ARIMA modeling
VRM	Compute Variance Reduction Matrix.
SMP	Compute Standard Deviation-Mean Plot.
ACF	Compute Auto Correlation Function for K timelags.
ACF(d,D)	Compute Auto Correlation Function for various degrees of non-seasonal differencing (d) and seasonal differencing (D).
PACF	Compute Partial Auto Correlation Function for K timelags.
Spectrum	Compute Normalized Cumulative Periodogram and Spectrum.
Spectrum/(P)ACF	Compute the Auto Correlation Function, the Partial Auto Correlation Function, and the Normalized Cumulative Periodogram (Spectrum) for the time series under investigation.
Estimate	Compute (estimate) all ARMA parameters (Full Information Maximum Likelihood Estimation).
Forecast	Compute the Univariate Stochastic ARMA Forecast (Ceteris Paribus Forecast).

Simulation of Profit Densities
PDensity	Compute the Profit Density of profits based upon a distribution-free simulation with 10*K = number of simulated profits, and M = 0 (default) for rejection of negative prices ('Metropolis' step). Note: do not set the number of K above 100 (= 1000 simulated profits) to save system resources.

General Descriptive Statistics
Timeplot	Compute basic statistics and chart of time series values.
Histogram	Compute histogram of time series values.
Rootogram	Compute suspended rootogram display of time series values.
Central Tendency	Compute various types of averages.
Concentration	Compute various types of concentration measures.
Moments	Compute centered and uncentered moments.
Skewness/Kurtosis	Compute and test various measures of Skewness and Kurtosis (small and large sample tests against normal distribution).
Quartiles	Compute Quartiles based upon 8 different definitions.
Percentiles	Compute Percentiles based upon 8 different definitions.
Variability	Compute various measures of Variability for the time series.

Series types
original series	Apply analysis to the original (raw) time series after Box-Cox transform and differencing.
residual series	Apply analysis to the residuals of the ARMA model (void if no ARMA model has been estimated).
squared residuals	Apply analysis to the squared residuals of the ARMA model (void if no ARMA model has been estimated).
interpolation	Apply analysis to the interpolation forecast of the ARIMA model (void if no ARIMA model has been estimated).
stationary variance	Apply analysis to W[t] = (lnY[t] - lnY[t-1]) (lnY[t] - lnY[t-1]) (this is used in the analysis of financial time series).
stationary range	Apply analysis to W[t] = \|lnY[t] - lnY[t-1]\| (this is used in the analysis of financial time series).
PD Buy and Hold	Apply analysis to the simulated Profit Density (PD) for the Buy&Hold strategy. Note: this assumes that the simulations have been previously computed.
PD Filter-rule	Apply analysis to the simulated Profit Density (PD) for the Filter-rule strategy. Note: this assumes that the simulations have been previously computed.
PD BH - Filter	Apply analysis to the simulated difference in Profit Density (PD): Buy&Hold profits - Filter-rule profits. Note: this assumes that the simulations have been previously computed.

Parameter list
L	Lambda: parameter of the Box-Cox transformation function.
d	Degree of non-seasonal differencing.
D	Degree of seasonal differencing.
s	Seasonality: number of observations per year.
K	Number of timelags to be used in computations (number of iterations in simulations = 10 K).
p	Degree of non-seasonal AR(p) polynomial.
q	Degree of non-seasonal MA(q) polynomial.
P	Degree of seasonal SAR(P) polynomial.
Q	Degree of seasonal SMA(Q) polynomial.
M	set M = 1 to include a constant term (M = 0 otherwise)

::Free Statistics and Forecasting Software::

:: Wessa.net - Web-enabled scientific services & applications ::