Hi there,

I got this EA result from a friend who has developed it. The results look great but how trust worthy is it? Can someone comment on it?

Attached Images

snapshot.jpg (183.5 KB, 167 views)

Go to:
Great EA in backtest!

This thing made almost 1500 trades over the course of 13 months and all you got was a 60% return? It may be profitable, but I think there are better systems out there.

First, it should be 90%.

And not for one year. Because market is changing. And EA should accept it. When we backtest EA we may know the following:
2001 - small profit in ...;
2002 - losses ...;
2003 - profit in ...;
2004 - losses in ...;
2005 - profit ...;
2006 ...

And EA may be improved according to these data.
Because when we backtesting so it is past data.
Well, we may make forward test. But when we finish this forward testing it will be the past data as well for us. But market is changing all the time.
So, if we are doing backtesting so it is important to do it for as many years as possible.

Or just forget about this backtesting at all and make some manual testing on indicators finding the settings and starting with forward test only. It is the other way which I like more than any backtesting.

Thanks for you comment. What i was focusing on is the drawdown vs. the profit ratio. Its drawdown is $1000 and its ROI is $6000 - it is risking $1 for every $6 that you may gain.

Hi Newdigital,

Since you are the foremost expert in EA testings, I have several questions for you:

1) what is the difference between modelling quality of 89.99% vs 90%?

2) How accurate is the backtesting program? If I were going back to manually check the 5min charts over the period of 13months (note that this EA trades on the 1hour chart), would i get 100% correlation between the manual verification versus the trade report generated by the strategy tester?

3) if the market is constantly changing with time, going to backtest over the period of the last 5-6 year and optimizing the EA for it, what would that gain us overall? Since the market is constantly changing, would it better for us to take samples of the last 200-300 trades and constantly adjusting the EA to adapt with the changing market i.e. creating a dymanic EA or AI.

4) How long would one test their EAs live before putting some money behind it, a month, two months, a year or two years? I am asking this question out of lack of experience, not to question your input.

5) There is a comment - "Good traders are not programmers and programmers are not good traders else all the good programmers are millionaires with their complex EAs". What is your take on this? I personally believe that we should all learn to trade and be profitable first before trying to program. Its GIGO - garbage in garbage out, if we can't trade profitable manual; how can we program profitable EAs? I know that I have upset many programmers with this comment but let's hear it from the other side.

It is some data not enough in strategy tester. Results may be diferent.

Sometimes yes, sometimes not. ersonally I don't like backtesting in strategy tester at all. But I am using it sometimes.

There are many theories how to backtest. I forgot the name of the articles. It is more related for optimization. And if we are posting EA for free - so everybody may do anything.

But if "somebody" is trying to sell EA based on good backtesting results only for example - it is the other case. Everybody knows everything in this case and we all are informed. And this "somebody" knows, and people who are buying know. But anyway, people are selling and buying based on backtesting and we can not do anything. All we can do is to say something about backtesting and about how to do it.

It's depends on the strategy. The best way is to trade the system manually for 2 or 3 months. And alway estimate the risk. Don't look at the good trades so much. Look/analyze the lossing trades to understand the risk. And I've never used EA in real account if I don't understand how this EA is working.

When we finish forward testing so it became a past data as well. The best way is to understand trading strategy, trade manually for some months to know how particular EA is working. But problem is that we all (including me) are having such a big expectation to make money using EA so we may forget about anything sometimes.

There are trading system developers and there are forex programmers. Different specialities. Programmers may code as coding machine. As a computer. If the programmer is improving the system according to some forex knowledge - so he is not programmer only. He is programmer and system developer. Not every programmer is system developer and not every system developer can program.

It is exceptional case when programmer and system developer is one person. When programmer and developer are working all together and understand each other so they may achieve great success.