Commands of SCA useful for Analysis of Financial Time Series:
Comment starts with the character ``--''.
Optional commands are in square brackets.

-- to save output into a file called ``scaoutp.otp''
profile review. [owidth 80]

-- to load a single column of data from file ``fname.dat'' into SCA workspace
-- Below xxx denotes any variable name of your choice.
input xxx. file 'fname.dat'

-- to load multiple columns (say 3) from file ``fname.dat'' into SCA
-- Format is the usual FORTRAN format.
-- Variable names x1,x2,x3 are selected by user.
input x1,x2,x3. file 'fname.dat'. [format '(4x,3f10.4)']

-- to obtain descriptive statistics of variable x1
desc x1

-- to quit SCA
stop

-- to stop a command due to typos (but remain in SCA)
quit

-- to compuate ACF of variable x1. The number of lags canbe chosen.
axf x1. [maxlag 12]

-- to compute PACF of variable x1
pacf x1. [maxlag 12]

-- to compute ACF and PACF of variable x1
iden x1. [maxlag 12]

-- to compute EACF of variable x1
eacf x1.

-- to do an automatic ARMA model selection for variable x1
-- If the series is seasonal, give the periodicity xxx.
iarima x1. [seas xxx]

-- to specify an ARMA(2,1) model and call the model model1 for variable x1.
-- numerical numbers inside the parentheses in front of x1 denotes the 
-- autoregressive lags used. No parentheses are needed if no AR part.
-- The existence of c1 means the model contains a constant term.
-- numerical numbers inside the parentheses before noise denotes the 
-- moving-average lags used. 
-- noise means the shock or innovation in SCA
tsm model1. model (1,2)x1=c1+(1)noise.

-- to estimate the specified model ``model1'' and hold residuals in ``r1''
estim model1. hold resi(r1). [method exact]

-- to produce forecast using model ``model1'' at forecast orign nn.
-- ``nofs'' denotes the number of forecasts needed, i.e. forecast horizon.
-- default nofs is 24.
-- default forecast origin is the end of data points.
fore model1. [nofs xx.] [orig nn] 

-- to obtain the roots of a fitted ARMA model ``model1''.
tsm model1. roots.

-- natural log transformation y = ln(x). 
y=ln(x)

-- Basically, SCA allows arithematic operations as usual, e.g. 
-- transform a log return in percentage into a simple return in percentage
y=100*(exp(x1/100)-1)

-- to obtain the squared series of a time series x1.
y=x1*x1

-- to obtain the absolute series of a time series x1.
y=abs(x1)

-- to obtain psi-weight and pi-weight of a fitted ARMA model
-- store the psi-weight in psi and the piweight in pi.
weight model1. psiweight psi. piweight pi. [cutoff 0.001]

-- to specify a regression model with time series errors
-- assume the regression is y=b0 + b1*x1 + b2*x2 and the time series 
-- error follows an ARMA(2,1) model.
-- call the model ``m2''. In front of the ``noise'',  numerator denotes
-- the moving-average part and denominator the autoregressive part.
-- If no MA model is needed, then use 1/(1,2)noise.
tsm m2. model y=b0 + (b1)x1 + (b2)x2 + (1)/(1,2)noise.

-- to compute correlations between variables x1,x2,x3.
corr x1, x2, x3.

Finally, once you exit SCA. The output file ``scaoutp.otp'' is 
in your directory. You may rename it if you like to keep it. 
On GSBMBA use the following command to rename it ``solut.otp''.

mv scaoutp.otp solu1.otp