Old Holy Grail?

 

I bactested my expert advisor which is based purely on price action strategy. It show great results between 1999 - 2007 but failed badly between 2007 - 2010. I tried & tested every year, capital size, lot size and other possible ways that I know but the result is still similiar. I've also tested it with other brokers and yes, same thing.


Could anyone explain what actually happened? I visually see that the candles on the chart between 1999-2007 is not that good or "broken". My question is this whole thing due to the tick data problem from Metaquotes or is simply just because my strategy did not work for the year 2007 - 2010?


Please excuse me if this is an irrelevant question as I am new to this backtesting thing.


General settings of my expert advisor is as follows;
Stop Loss: 50 Points
Take Profit: 10 Points
Currencies Tested: EURJPY, USDCHF, EURCHF, EURUSD, EURGBP
(similiar results on other currencies especially EUR)


Result from 01/01/1999 - 01/01/2011

 

This probably is a mixture of spread-sensitivity, and maco changes in the market

There were 3 major market pattern changes in 2007, 4 in 2008, more in 2009 & 2010

Your system doesn't seem suited to any of the more recent ones

Many long-term profitable EA's stopped working mid-2009

FWIW

-BB-

 

that reminds me on a old project: https://www.mql5.com/en/forum/126819

Seems that your strategy need a low ATR to be successful.

You see the changes in price action if you open a daily chart. The market moves get more agressive (fast) and the average range increased a lot.

//z

 
Invalid:

I bactested my expert advisor which is based purely on price action strategy. It show great results between 1999 - 2007 but failed badly between 2007 - 2010. I tried & tested every year, capital size, lot size and other possible ways that I know but the result is still similiar. I also tested it with other brokers and yes, same thing.


Could anyone explain what actually happened? I visually see that the candles on the chart between 1999-2007 is not that good or "broken". My question is this whole thing due to the tick data problem from Metaquotes or is simply just because my strategy did not work for the year 2007 - 2010?


Please excuse me if this is an irrelevant question as I am new to this backtesting thing.


General settings of my expert advisor is as follows;
Stop Loss: 50 Points
Take Profit: 10 Points
Currencies Tested: EURJPY, USDCHF, EURCHF, EURUSD, EURGBP
(similiar results on other currencies especially EUR)


Result from 01/01/1999 - 01/01/2011


because every day/week/year in the market is different :-)

So - sometimes will work, sometimes not.

But maybe work better if you dont use fixed SL and TP.

Look for other ways, like Values of StdDev or ATR to use for this (ONLY EXAMPLE!!!!)

SL = 4*StdDev, TP = 2*StdDev

 
EADeveloper:


because every day/week/year in the market is different :-)

So - sometimes will work, sometimes not.

But maybe work better if you dont use fixed SL and TP.

Look for other ways, like Values of StdDev or ATR to use for this (ONLY EXAMPLE!!!!)

SL = 4*StdDev, TP = 2*StdDev


It’s a great Idea. It really makes sense. Market adaptable.

 
BarrowBoy:

This probably is a mixture of spread-sensitivity, and maco changes in the market

There were 3 major market pattern changes in 2007, 4 in 2008, more in 2009 & 2010

Your system doesn't seem suited to any of the more recent ones

Many long-term profitable EA's stopped working mid-2009

FWIW

-BB-


you mean scalping EA's right ?

 
zzuegg:

that reminds me on a old project: https://www.mql5.com/en/forum/126819

Seems that your strategy need a low ATR to be successful.

You see the changes in price action if you open a daily chart. The market moves get more agressive (fast) and the average range increased a lot.

//z

Too bad, I didn't come across your old project, it's so similiar. And yes, I tried ATR before coming here and the results are still not good. :-(

Did you succeed fixing your old project?

BarrowBoy:

This probably is a mixture of spread-sensitivity, and maco changes in the market

There were 3 major market pattern changes in 2007, 4 in 2008, more in 2009 & 2010

Your system doesn't seem suited to any of the more recent ones

Many long-term profitable EA's stopped working mid-2009

FWIW

-BB-


You're right, the major market pattern changes within the years and that left me with no option but to abort trying this strategy. Thanks, that make sense to me now.
 

I have many different strategies with same result, i think if you test any strategie with small takeprofit from 1999 till 2007 it give you good result becaue the moves from the price where other in the past, the candels where longer, now today the candels are most time smaller and just some singel candels move fast and long, so scalping is then not so good possible as before with the EA strategies.

And if you look today to AUDNZD and simular pairs you can see the candels are simular like EURUSD and other majors pairs before 2007, but AUDNZD have much spread and in the past the spread for the major pairs where also bigger then today.


if now the spread for AUDNZD become smaller then the character from that pair will also change, it will look then more simular like EURUSD today looks.