Answer: Simetar© is compatible with Windows® 95, 98, 2000 or XP operating systems and Microsoft® Excel 97, 2000, 2002, 2003 or XP. The only restriction for running Excel 95 and 98 is that the computer must have a full install of Excel including the Analysis Pack and Solver.

Answer: Simetar© is a complete risk analysis software program that contains tools for estimating parameters of probability distributions, statistical analyses, multiple regression, forecasting, random number generation, ranking risky alternatives, and graphing and presenting simulation results. Additionally Simetar© has an option to incorporate Solver into stochastic analyses so simulation models can be optimized during simulation.

Answer:

Simetar© commands look like Excel commands and can utilize the Excel equation editor dialogue box, making it easy for users to quickly program Simetar’s most complex functions. Commands for generating a random number from Uniform and Normal distributions in Simetar© and Risk are provided so Risk users can see how easy it is to change to Simetar©.

Distribution
Normal
Uniform
Simetar© Command
=NORM(mean, std dev)
=UNIFORM(min, max)
@Risk Command
=RISKNORMAL(mean, std dev)
=RISKUNIFORM(min, max)

Answer: Help is available for all Simetar© functions by selecting the help icon and searching the function menu. Extensive explanations and examples are provided for each Simetar© function. References are provided for most functions if the user wants additional information for a function.

Answer: A complete Users Manual comes with Simetar©. Simetar© training workshops are provided to help users get the most from the software. Help is provided in Simetar© through documentation of all functions and menus available to the analyst. If users have specific problems or questions, they are answered, via the internet, usually within 24 hours.

Answer: If your macros are written to execute when a particular cell changes and the change cell is either directly or indirectly associated with a random variable, then the macro will run during a Simetar© simulation.

Answer: Simetar© can simulate multivariate normal and several non-normal probability distributions. Simetar© uses the standard correlation matrix (or the rank correlation matrix) to simulate correlated uniform standard deviates which are then used in the inverse transform of any of the 40+ probability distributions supported by Simetar© to simulate correlated non-normal random variables. For example, a multivariate normal variable is simulated with =NORM(mean, std dev, CUSD) where CUSD is a correlated uniform standard deviate calculated with Simetar’s =CUSD(correlation matrix) array function. This procedure is one method to generate a multivariate normal copula.

Answer: Simetar© includes five univariate tests for normality and a large sample test for multivariate normality that can be applied to any data series. A univariate parameter estimation system is included which estimates parameters for 16 parametric distributions using method of moments (MOM) and maximum likelihood estimator (MLE) procedures. Equations for simulating the 16 distributions are generated so the user can simulate the random variable for each distribution and compare the simulated data to the historical series. Three estimation options are provided for estimating parameters for empirical probability distributions.

Answer: No. The multiple regression function in Simetar© allows the user to specify models with up to 250 exogenous or right hand side variables. The limit of 250 comes from the limit on the number of columns for displaying the results in an Excel worksheet.

Answer: For Simetar© 2005 the number of observations is limited by the size of the Excel spreadsheet.

Answer: Simetar© gives the analyst the option to generate random numbers from four different random number generators: Mersenne Twister, L’Ecuyer’s Combined MRG , Excel’s native generator or the C++ random number generator. The analyst can specify which random number generator to use or use the default generator, the Mercene Twister.

Answer: Excel’s native random number generator can be selected, but the analyst may elect to use the Mersenne Twister or L’Ecuyer’s Combined MRG or the C++ random number generator.

Answer: Yes Simetar© defaults to Latin hypercube sampling for each of the three random number generators. However, the analyst may elect to sample random numbers using the Monte Carlo method.

Answer: Yes the user can set the random number seed for the Mersenne Twister the Native Excel and C++ random number generators. L’Ecuyer’s Combined MRG method uses 6 seeds, and Simetar© allows the user to select which set of seeds to use from 100 sets of pre-selected seeds.

Answer: The maximum number of iterations is dependent on the number of rows in a worksheet (65,536) and the number of output variables for which statistics are calculated. If the number of output variables is less than 255 then the maximum number of iterations is 65,525. If the number of output variables is between 255 and 508 the maximum number of iterations is 32,750.

Answer: The number of output variables for which Simetar© calculates statistics at the end of the simulation is virtually unlimited. If the analyst specifies more than 254 output variables, Simetar© puts the additional variables below the iteration results for the first group of 254. If the model is simulated 2000 iterations the maximum number of output variables is 8,100.

Answer: Yes. During simulation Simetar© can access Excel’s Solver add-in to optimize the simulation model. This option is particularly useful for simulating a simultaneous equation model (such as, supply and demand) that has stochastic inputs.

Answer: Simetar© utilizes Excel’s Solver to sim/solve a stochastic simulation model. If your problem will optimize deterministically with Solver then it generally will solve the problem during a stochastic simulation in Simetar©.

Answer: Simetar© contains several menu driven statistical tests for performing validation tests of random variables. In the case of univariate distributions, the Student’s-t test and the F test can be used to test if the simulated mean and variance equal the historical values. In the case of multivariate distributions, the Hotelling T2 Test is available to test if the vector of simulated means equals the vector of historical means. Box’s M Test is available to test if the covariance for the simulated data equals the historical covariance. The Complete Homogeneity Test is available to test if the means vector and covariance are simultaneously equal to their historical counterparts. A Student’s-t test is also available to test if the simulated data have the same correlation matrix as the historical data for a multivariate distribution.

Answer: No additional software is needed. Simetar© contains a comprehensive multiple regression function which estimates the parameters for a regression model, as well as forecasts the dependent variable for as many periods as exogenous values are provided and calculates the forecast standard deviation for each forecast period. Simulating a probabilistic forecast in Simetar© is very easy, given that the forecasted means and standard deviations are provided and they can be cell referenced to the =NORM(forecast meant, forecast std devt) formula without a loss of accuracy due to importing these values from another software package.

Answer: Yes. Bootstrap regression for cross sectional (sample) and time dependent exogenous variables is very easy to do in Simetar©?
Confidence intervals for regression coefficients can be calculated and displayed graphically in a probability density function.

Answer: Yes. Jackknife estimates can actually be done for any cell in an Excel spreadsheet, not just the mean and variance for a variable. For example, jackknife estimates can be calculated for any statistic associate with a vector or matrix of data.

Answer: Yes all output reports, charts, and risk ranking summaries that are cell referenced to the simulated values are automatically updated each time the workbook is simulated using the same output variables.

Answer: Simetar© provides dialogue menus so the analyst can develop charts of the simulation results. The charts included in Simetar© are: line graph, probability density function (PDF), cumulative distribution function (CDF), fan graph, Stop Light graph, stochastic efficiency with respect to a function (SERF), risk premium between risk alternatives, histogram (as a PDF or CDF), probability plots (Normal Probability plot, Quantile-Quantile plot, and Probability-Probability plot), Box plots, and Scatter Matrices. The PDF charts can be drawn using alternative kernel estimators, such as Gaussian, Epanechnikov, Parzen, etc. and the confidence intervals for the PDFs can be displayed. The CDF charts also can be smoothed with alternative kernel estimators or presented in their empirical form.

Answer: Simetar’s graphics capabilities can be used to develop charts of any series of data, whether it is historical data or simulated data. Additionally, Simetar© can develop a chart which compares the historical series to the simulated data on the same axis. For example, the 20 years of historical data for a random variable can be displayed with the simulated values for the same variable as a PDF or a CDF.

Answer: Simetar© charts are actually Excel charts so they can be edited and modified using any of Excel’s functions for changing the type, format, colors, etc. of a chart.

Answer: Simetar© charts are Excel charts that can be copied and pasted special as Meta files into any Microsoft document, such as Word or PowerPoint.

Answer:

Simetar© has an option to simulate a sensitivity analysis of the effects of many control variables on a key output variable and then present the results in terms of sensitivity elasticities in both a table and a chart. Sensitivity elasticities quantify the percentage change in the key output variable with respect to a one percent change in a control variable. Simetar© also has an option to conduct a sensitivity simulation analysis of a spreadsheet model with one control variable being set to six different values. The results of the analysis can be used by the analyst to construct a chart.

Answer: The dialog menus in Simetar© are only in English, but the function and menus work in many different languages. Simetar© runs successfully on computers operating in Spanish, Norwegian, Swedish, Finish, Dutch, German, French, Italian, Korean, Vietnamese, and Chinese.