Looking for experienced traders/coders to explain backtest result

I need some help regarding an EA.

I've done a massive amount of backtests, and all of them are very good. How come my live tests arent giving me same results?

There must be some good explenation, because these backtests are crazy...

This one is from 06-11-2014 till now

Maybe post a comparison of the back- and forward-test for the same period? I think that would help to diagnose the problem.

Probably these are factors:

1) Slippage
2) Spread widening on news
3) If your EA is making trades off of current bar (0), it may get more false signals as the current bar is moving. This may generate a signal, which is later not a true signal. Try using previous bar (1) which is completed for signals.

Regards,
EZ

Dear Ullen

Lots of factors has to be taken into consideration. Are you taking the trades as per you plan, are you skipping any trades. Are you trading the same pairs as your backtest ?

Also, how many forward trades have you taken, if backtest results are for 10 years and you only forward testing one month, its not enough data..

Hope it helps

Backtests are bullshit.

If they worked, how many EA zillionaires would be out there?

Would anybody trade manually?

Keep with forward testing and test multiple brokers, this is the only "True" data that you will get.

Also note that in the backtests there are periods of significant length where ea is losing money. If your forward tests were of short period compared to the backtest time, it could also be that it is coincidental, i. e. that the current market phase does not align well with your ea's strategy.

k

Thought about this, but this one is also over a 3 months period of time.

Seems too good to be true

Hello, Ullen,

I am experienced coder and unfortunately know few issues with MT4 strategy tester... My advice - don't believe in tester result. You didn't tell us much about your system, what I understand, but is quite hard to help you this way.
To be specific:
- great advice from ezcurrency -> use bar[1] instead of bar[0]
- do forward test a lot and believe ONLY on that results (if your system is really that good you don't need to rush, open demo account and let it run)

Will ad more infos about strategy tester when I will find links.


All the best to you.

...here is material I promised:
https://www.mql5.com/en/forum/13542 - it is really good documentation about MT4/5 strategy tester

I'm very new to this EA world. Can you explain how i use this ''bar[1] instead of bar[0]'' ??

Thx

Bar[1] is the value at close of the last bar.
Bar[0] is the current candle.

So, it's like this...

If (stochs[2]<stochs[1]&&stochs[0]>stochs[1]) return (-1);

Assume stochs is the numerical value of either the main or trigger line on the stochs indicator...

It is saying that if the stochs previous was less than what it was 2 bars ago and it is currently greater than the previous bar, buy...

What happens is, the value of the indicator and price fluctuate at [0] value because of the movement of the bar. IF you set these as a delay to value [1], the EA is forced to wait for a single bar to be closed on value prior to triggering.

All that being said...

Backtests are bullshit and this will not give you results that will match live forward testing, but it can stop an EA from taking false triggers in a whipsaw market.

I totally agree with Magix! You need to forward test in demo to see how it does for several months in different market conditions (and on various currency pairs). If you are happy with the results, then put in a little bit of $ to trade ... if it is still doing well 3-6 months later, then put in more.

Regards,
EZ

Than you guys.

Forward test is still going on, and good results are showing

I have done massive tests on strategy tester on MT4 on many trading systems. I give you some tips.

1. check the profit factor on report. It must above 8, then, your EA may be ok on living environment for long-run. So, most of EA fail with near breakeven profit factor, which is why most of human traders or EA traders lost out sadly. Too many of "fx experts, tutors, trainers, gurus are wrong with their trading systems and their teaching/training".

2. check the report, on largest profit trade, largest loss trade, average profit trade, average loss trade. A sustainable profitable trading system must have this feature:
(1) average profit trade must be significantly greater than largest loss trade.
(2) average profit trade must be significantly greater than average loss trade.
(3) win % must be greater than 50%.

These benchmarks are for any good EA, otherwise, EA will fail out in living trading. Check these (Jack1) benchmarks against your EA, then, you will understand better of your EA.