Backtest history vs live history

New comment
 
Hello everyone, 

I've built an EA and have been running it on a VPS for a week. In the meantime I made several tests with the tester of MT4 to see how my EA behaves with different pairs and also with the different time frames. 

However, I have noticed that the tests are very different. My EA made profit on the live servers but it made a small loss while backtesting it in MT4. 

Has anyone of you ever noticed this or is this a known problem? The EAs run with the same settings and I've also taken the news filter into account. 

It also places different trades. I'm just curious if I'm doing anything wrong and if should even put a little bit trust in the backtests I made.

Thank you for the answers and best regards,

Lucas

 

  1. Why did you post your MT4 question in the MT5 EA section instead of the MQL4 section, (bottom of the Root page)?
              General rules and best pratices of the Forum. - General - MQL5 programming forum? (2017)
    Next time, post in the correct place. The moderators will likely move this thread there soon.

  2. It has not “run with the same settings.” The spread in the tester is constant during the run. The market is not. Did you run with a valid average spread to even get close?
  3. If “it also places different trades,” your code is suspect.
 

1. I pressed on forum and then on "new topic" next to "Expert Advisors and Automated Trading". I didn't know that there is an extra topic for MT4 since I didn't scroll all the way down.
I'll take a better look the next time. Thank you for the info. 

2. My EA has a Spread filter. I asked my Broker for the Spread history and they gave it to me for some pairs. That's the pairs I've tested so far. I imported the list after each test in excel and deleted all the trades that my EA wouln't have taken anyway since the spreaed was above the limit. I set the spread in the tester to the max my EA would still take and let it run on that. So in theory  the EA did worse on the test then in real? I also used a fixed lot and calulated the average win and loss in money and pips/points. 

3. What exactly do you mean by suspect? My problem is that it places the same trades on live and demo, just not in the strategy tester. It also places the same trades on my live account and on the one from a friend. That's why I'm asking if I'm doing something wrong with the backtest in general. It is actually placing the right trades on the live accounts but skips on profitable trades in the tester.

Can it be due to my Chart history? Sadly I don't have any history of other EAs to test if my History is the reason.

 
Hi, if you have poor quality data, your back test will have poor quality, in results, your back test could be a fiction. Regards Greg
 
Grzegorz Pawlak #:
Hi, if you have poor quality data, your back test will have poor quality, in results, your back test could be a fiction. Regards Greg

Hey, 

that's totally right. I'm using Tickstory since it seems to be quite reliable and known with some nice features. 
The Statement at the end shows a 99.90% Quality. I also talked to the Tickstory support if I did something wrong but according to them that shouldn't be the case. 

It's nice that you point it out and I should've written it in the last comment but I've already taken care of it. 

New comment