Thread: Why doesnt it work more recently?

Hi All,

I have been reading as far as I can from the forum, and learning heaps. Thanks for everyone that leads. I even then went out and heard MQL (already had C experience, so was fairly easy). But I am having one strange issue. I composed an EA, and once I back analyzed, it worked quite well... kind of. By May 2003 (where I started the testing) up until October 2006, it consistently made a very good profit. Since it made a reduction. The turn was too much to be a coincidence.

Another thing that may or may not be worth noting, so I'll say it anyhow... when running long evaluations, it appeared to just stop trading after a particular amount of time. I assume you will find a limit to the number of transactions it can capture?

I imagine the first issue will be where did I get my data? The answer is that I went to tool, history centre, and downloaded it (clicked download).

Has anybody had this problem before? Does anybody have any suggestions?

Oh, incase it's relevant, it had been EURUSD, 30M, and the version was each tick.

In case a lot size increases account balance your EA will choke after it matches its maximum lot dimensions ( in many cases 100 lots, varing by broker ) this occurred to me while backtesting...I was baffeled for hours

Also MT$ is merely glitchy sometimes... reinstall the program might clear it up.

And the market has shifted drastically from 03 - 06.... Things which were true then are fairytales today...

You know this already, but backtesting in MT4 is notoriously unreliable. Have forward tested it in demo? I prefer tested resuts over back.

Thanks guys. I didn't know about the max lot size, that would explain why it ceased trading at a certain point (yes, I had been raising lot size). In terms of the market shifting, I don't have any doubt it has, but I don't think that's my problem here... it happened too abruptly. I will tell you the day where the machine moved from profitable to unprofitable.

As for MT4 back testing being undependable, how undependable. Have many individuals had strategies that were profitable back testing instead of forward testing? Or the reverse? As for me, I wouldn't have bothered forward testing if it does not pass back testing. Is that a mistake?

Originally Posted by ;

because men. I didn't know about the max lot dimensions, which would explain why it stopped trading in some stage (yes, I had been increasing lot size). As for the market shifting, I don't have any doubt it's, but I don't think that is my problem here... it happened too abruptly. I can tell you that the day where the system moved from profitable to unprofitable.

As for MT4 back studying being undependable, how undependable. Have many individuals had systems which were profitable back studying and not forward testing? Or the reverse? I would not have bothered forward whether it doesn't pass back testing testing. Is this a mistake?

That the backtester is totally undependable for intra-bar ticks as these ticks are system generated on which MT4 guesses may have happened. Then the MT4 backtester is going to do just fine, however if your EA just looks at the previous close of pubs for all purposes.

I also have seen the exact same thing, I think that it is the market going though phases, for example moving averages used to operate well but then the market became even more volitile, last year martingale EAs did well, yet this year. . .not so good as we appear to be in a phase of slow trending.

The increase in $$$$ on the place market in the last few years is huge.

Thanks aicccia, that is handy to know, and will most likely save me a lot of time later on.

Some individuals have mentioned that the market is changing, but surely the market changes fairly slowly, as in one month MAs work very nicely, then there is a transition period in which the work better some months then others, then gradually, they do not really work well whatsoever. (I'm open to being corrected if this is not the case). This is not what happened in my EA, it was literally profitable most weeks up until a particular date, then was very seldom profitable.

Having said all this, it's a bit of a mute point given aicccia's inputsignal, since it did rely on ticks... but I'm just worried that my data is dodgy, and mystified why it happened.

What date does this turn on you?