thankfull holy grail

[Gabriel Mejia]

I present you guys.... the holy grail......tataaaaa

no...just joking. this is a crappy EA that converts 900 USD into 4.000.000 USD in about 5 years in testing. EURUSD 5 min.

its a Martingale EA. 

Obviously  I know that this is imposible.  And test results dosen´t mean anything.   forward testing with a live account is needed.  But as an intelectual exercise I want to understand the reason it produces those results on backtesting. I want to understand what is the mistake done by the code or by metatrader history data  that produce this kind of results. This EA works on EURUSD 5 minutes everythick. Can you guys please educate me where is the catch. This code was writen by me but those are not the results I was expecting. As a matter of fact, I was just doing a intelectua excecise when I writed the code.  

The history data was dawnloaded from http://www.histdata.com/download-free-forex-data/. I downloaded de 1M information for EURUSD (5 years of data) since 2010 and then used the periodconverter script to generate all the timeframes (M1,M5,M30,H4)

 

Someone can help me to find the error and how is it generated?.

 

(I suspect is a problem with the history data) 

                                                                                                                          

                                       

                                                                       Strategy Tester Report

thankfull

ETXCapital-Live2 Server (Build 910)

Símbolo		EURUSD (Euro vs US Dollar)
Período		4 horas (H4) 2010.10.06 00:00 - 2015.11.27 20:00 (2010.10.06 - 2015.11.28)
Modelo		Cada tick (el método más preciso basado en todos los períodos menores disponibles)
Parámetros		precio_entrada_largo=0; precio_entrada_corto=0; volatilidad=1; riesgo=3; max_perdidas_consecutivas=1; val=101; alturaMacd=0.0004; promedio_ganancia_macd=0; activatendencia=1; tipo_metodo="Metodo_convencional"; tipo_sistema="Directo"; distancia=110; s1=9; s2=3; s3=3; m1=1642; m2=50; metodo=0;
Barras en la prueba	7365	Ticks modelados	28234174	Calidad del modelado	n/a
Errores de gráficos mal agrupados	666247
Depósito inicial	600.00			Diferencial	16
Beneficio neto total	317964.93	Beneficio bruto	341797.33	Pérdida bruta	-23832.40
Factor de beneficio	14.34	Rentabilidad esperada	27.62
Disminución absoluta	0.00	Disminución maximal	5080.20 (2.30%)	Disminución relativa	11.07% (1382.57)
Total de operaciones	11512	Posiciones cortas (ganado %)	11478 (99.77%)	Posiciones largas (ganado %)	34 (55.88%)
		Operaciones de beneficios (% del total)	11471 (99.64%)	Operaciones de pérdidas (% del total)	41 (0.36%)
Mayor		Operaciones de beneficios	2038.56	Operaciones de pérdidas	-2936.00
Media		Operaciones de beneficios	29.80	Operaciones de pérdidas	-581.28
Máximo		ganancias consecutivas (beneficios en dinero)	2090 (46549.95)	pérdidas consecutivas (pérdidas en dinero)	5 (-1687.55)
Máximo		beneficios consecutivos (número de ganancias)	52640.00 (672)	pérdidas consecutivas (número de pérdidas)	-2936.00 (1)
Media		ganancias consecutivas	441	pérdidas consecutivas	2

[Carl Schreiber]

ahem - were you asking what is the mistake because your EA is so successful?

In your case it might just because you have chosen a 'lucky' data window!

Do you see the first big volume peak between 3882 and 4360? Start your EA right before that peak! I guess your account will be wiped out!

One advice if you close a bunch of orders one by one - you HAVE to count down!

// wrong way: for(cnt6 = 0; cnt6 < abiertas_este_activo6; cnt6++) {
for(cnt6 = abiertas_este_activo6-1; cnt6>=0 ; cnt6--) {
   if ( !OrderSelect(cnt6, SELECT_BY_POS)) { Alert("???"); continue; } // check the outcome of the functions!!
   if(OrderType() == OP_SELL) {
        Alert("cerrando 3");
        OrderClose(OrderTicket(),OrderLots(),Ask,3,Violet);
   }
   if(OrderType() == OP_BUY){ 
        Alert("cerrando 4");
        OrderClose(OrderTicket(),OrderLots(),Bid,3,Violet);
   } 
}

BTW for historic data you can use "Tickstory Lite" or Birt's "Earview" sccript or "Strategy Quant's Tickdownloader" (google)

[ffoorr]

I think you did not made a each tick backtest, on M5

7365    Ticks modelados 28234174        Calidad del modelado    n/a

Período4 horas (H4) 

It would have been too long to backtest several years tick  by tick

The tick by tick backtest is usefull for scalper, but for other stategies like M5 strategie, M1 Bars are ok, and the data from  metatrader are sufficient.

 

edit :

ok this is the problem, i don't know what's the origin of it

Errores de gráficos mal agrupados       666247

[Doerk Hilger]

Normally an unadjusted martingale does not even survive a year, assuming you´re doubling at every percent of the underlying symbol, e.g. in EURUSD you´d double on every cent of the EUR the last position size - which means you have a progression factor of 2.0, that's real martingale then.

I am curious, which settings you have used? Distance, take profit etc.?

Doerk