Bad Data Filtering

["Mark Brown" <markbrown@xxxxxxxxxxxxx>]

the problem with filtering bad data is not how to filter it or what to
replace it with.  the big mystery is how to avoid the anomoly data from
having input into future calculations.

========================


input : vbars(10),shockfac(3),endshock(1.00) ;
vars : vlty(0),norm(0),avgTr(0),ix(0),iy(0),na(0),vmax(100),trend(0),save(0)
;
arrays : rng[100](0) ;
{average true range during selected market periods}
if currentbar <= vbars then begin
{initilization}
if na <= vmax then begin
na = na + 1 ;
rng[na] = @truehigh - @truelow ;
end ;
end
else if currentbar = vbars + 1 then begin
{sort range and find initial median value}
for ix = 1 to na - 1 begin
for iy = 2 to na begin
if rng[ix] < rng[iy ] then begin
save = rng[ix] ;
rng[ix ]= rng[iy] ;
rng[iy] = save ;
end ;
end ;
end ;
avgTr = rng[@intportion(na/2)] ;
end
else if currentbar > vbars + 1 then begin
{normal processing}
{current volatility uses all bars}
vlty = @truehigh - @truelow ;
{test for price shock levels}
if vlty/avgTR > shockfac then begin
if vlty>shockfac then
norm = avgTR ;

if vlty>=0 and vlty<=1 then plot1(vlty,">shock1");
if vlty>=1 and vlty<=2 then plot2(vlty,">shock2");
if vlty>=3 and vlty<=4 then plot3(vlty,">shock3");
if vlty>=4 and vlty<=5 then plot4(vlty,">shock4");

end ;end;