Expert Advisor Advice for MT4

Trading Systems Expert Advisors and Automated Trading
1

Hi guys a question, i have a new EA that i have spent about a month tweaking the settings (as the default settings failed misrembly) and for the last few days have had great results testing on a demo account.

Just so you know i am only running the EA on EUR/USD between the hours of 0700 - 1600 GMT UK time. After 1600 i will let the last trades finish. I think this is best to capture the best hours of trading this pair when there is most movement.

But i have a few questions as i am new to EA’s. I know the basics such as testing, tweaking and testing and don’t trust them ! lol. I come from a technical IT background so i am familiar with testing and the reliability of software and hardware, i just don’t want to get it wrong.

  1. How long do you test an EA on a demo account (providing it gives good results before going live) ?
  2. What are your success criteria at the end of the demo period (as defined in question one) before you either bin the EA or tweak it more (i believe i have the optimum settings now, but further testing is needed to prove me right).
  3. I have heard the backtesting in MT4 is unreliable compared to running in a demo account, but how reliable is testing in a demo account compared to running in a live account and what should should i be looking for in the demo account to give me signs this EA would be success or failure in a live account?

Thanks guys.

2

Backtesting with MT4 is not reliable because maximum modelling quality that can be achieved is 90%. You can do 99% modelling quality backtest which are more reliable but the history of the 99% modelling quality is only 5 years old. Well… Almost 5 years. (info about how to backtest with 99% accuracy here: eareview.net)
Before going live with real koney I would do 12 years backtest on MT4 90% to see if the account survives, followed by 5 years 99% to see how it performs in the last period with high accuracy, and if all is OK start the demo for at least 1 month. If demo performance is OK, go live with real money with minimum lots while keeping the demo running to check if the demo trades similar with the real account. When you gained enough confidence running live with minimum lots, go ahead and use the risk advice you got from the creator of the system, of course after you have checked other people’s real accounts and also and the most important the vendor’s real accounts. If they dont exist or there is one account funded with a silly amount of money, probably the system doesnt work and you dont have to bother taking all the steps described here.

Thats about all. Hope it helps.

Regards.

3

Thanks for the advice very helpful!!

Ill have to do more back testing later

So far over the last three days i have had 20 profitable trades at 10 pips each only one loss of 6 pips which was due to me following my rules and shutting down the ea at 1700 yesterday and only trading 0900 - 1700 uk gmt hours.

This EA seems promising and when i did one years of backtesting in MT4 i got a 10% return for the year.

Thanks for the advice!

4

Hi there,

  1. demo testing is good to get a sense of how an ea trades but it is basically worthless. Because live accounts are subject to re-quotes and delays, and demo accounts aren’t. Your ea may look successful in demo and fail miserably in a live account. The way to overcome this is to demo a little to get the sense of the ea than go live with a very small micro account and see how it performs in that environment.

  2. Success criteria can be measured in positive expectancy. you could have an ea lose more than 50% of the time and still have a positive expectancy if the risk:reward is great. or you can have an ea that wins 90% of the time and still have a negative expectancy because its stop loss is huge compared to its take profit and therefore its risk reward is way out of wack. At the end of the day you need to see that your ea is making money overall in terms of percent gain not necessarily pips.

  3. Backtesting is okay to get a sense of things like max drawdown and average win verses average loss but there is no crystal ball and things that look great in backtest often fail in forward testing. Its like the coin question…

If you take a coin and you flip it 7 times and every single time it lands on heads. On the 8th flip what are the chances it will land on heads again?

-dramatic pause-

still 50/50 your odds didn’t change. Its sort of the same with backtesting. It can give you the sense of the biggest string of losers or the max drawdown but at the end of the day the info is only useful for your overall analysis. It really boils down to positive expectancy and that can really only be determined through forward testing on a live account.

5

Thanks guys you both have been very helpful i was planning another week of demo testing and only use a small account with 1 micro lot as a max trade to see what happens

I am getting a feel with this EA but so far this EA has exceeded my expectations.

So hopefully nexts weeks of testing goes well!

I have had to play with testing this EA with alot of different settings to get it to where it is now, so all good so far.

If it works i’ll post the EA up with evidence so everyone can benefit only fair to share :slight_smile:

6

Looks like a lot of very good advice indeed. One question for Johnny though:

What about with Brokers that guarantee NO re-quotes, like ECN/STP Brokers, do you think the demo would still have no merit?

I know there are countless examples of how demo’s have not been an accurate reflection of live results, I just wonder how many of them were subject to what I mention.

Cheers

p.s BTW Sandy GL with the EA and testing, what is the core strategy?

7

Demos have almost no merit. Some times although, demo trading can coincide with live trading in case of certain strategies. For example strategies based on high time frames like H4 or D1, or strategies that use only candle close or cadle open prices shouldn’t differ a lot in demo enviroment. Keep in mind though that demo it just a DEMO. It is an ideal case. Other factors that could influence live enviroment are errors like old tick, off quote, slippage, broker disconnection issues, etc. So demo trading is just to make a general ideea about an EA.

Regards.

8

Thanks for all the advice guys, i really do appreciate it. It has made my expectations more realistic and will i think i know how i will use this EA.

With this EA i plan to use it as a manual guide. So far since Wednesday i have had 18 trades hit my take profit of 10 pips, one i close at -6.3 pips on thursday as i wanted to keep to my rules of finishing at the end of the day.

I also was not going to run the EA today (friday) due to all the advice from everyone to avoid trading friday afternoon but since its an EA and a demo i thought what the hell it will help me learn something and it did !!

With the Euro dropping currently as of now i have 4 trades, one at +1.7 pips, one at -18.3, one at -38.3 and the last at -58.3, so hopefully they will close with profit by monday.

The EA i think is good providing you have a human hand to watch it and close out bad trades and stop it when news comes which i did not do today !!! If i had stopped trading at 12.00 today like i wanted to i would of only have one active trades at -58.3 pips, so far so good.

Im not sure what you mean by GL ? but the EA is a grid based system which uses Stochastic, SAR and MACD.

I am not the creator of this EA and the only changes i have made is that i have amended the EA to my prefered settings and i have amended the credits to reflect the owner and where i got this script from to show i am not trying to take credit for someone elses work. Also i have renamed the main file to fit it with how i work.

If anyone has any suggestions as to how to improve the EA please feel free to make them and post the updated EA. I would like the ability to set a start and end time for each day and when that end time is reached the EA removes any pending orders (not executed) and will only run during the set times for each day.

Can anyone here do this?

Thanks Guys.

PS im thinking of testing the EA with a trailing stop for profits but i have not tested this yet, also i am away for the whole of next week so i cannot continue testing but i will leave the current active trades running to see what happens.

If anyone else tests this can we collaborate, perhaps someone can improve upon settings used that i have come up with?

Thanks all.

Sandy v1.00.zip (6.24 KB)

9

Just got the GL. So not used the internet speak lol :slight_smile: Cheers. Will keep you posted. If the EA works well in demo i will take it forward with a limited live account and see what happens and if it works i’ll let you all know :slight_smile: