Hmm... ok, I'll bite. Just to see... I put together this little test based on Workshop4_1:
Code:
#include <profile.c>

function run()
{
	set(PARAMETERS);
	
	vars Price = series(price());
	vars Trend = series(LowPass(Price,1000));
	
	Stop = optimize(4,1,10)*ATR(100);
	
	if(valley(Trend))
		enterLong();
	else if(peak(Trend))
		enterShort();
}


When I train this, Zorro assigns an optimize value of 5.09 in the .par file. If I TEST next, I get this:
Quote:
Read dt-test-par_EURUSD.par
Profit 5$ MI 0$ DD 21$ Capital 19$
Trades 441 Win 18% Avg +1.1p Bars 69
AR 5% PF 1.04 SR 0.07 UI 67.2% Error 41%


But then if I change that line in the script to read:
Code:
Stop = optimize(10,10,15)*ATR(100);


then without reTrain, only TEST gives this result:
Quote:
Read dt-test-par_EURUSD.par
Profit 5$ MI 0$ DD 21$ Capital 19$
Trades 441 Win 18% Avg +1.0p Bars 69
AR 5% PF 1.04 SR 0.07 UI 68.9% Error 41%


The difference is negligible, but why is it different at all?