Noob Question About BackTesting an EA

 

Hello,

I have been playing around with an EA I am built and am getting confusing results. I am using Tick by Tick Model Data. My EA trades on 15 min bars. When I backtest over 6 months data, I get excellent results, however it only shows modeling quality of 50.38%. When I backtest over 2 years, the code just doesnt consistently win until that last 6 months data. At first I just thought it was the market changing, however over the 2 years it showed 90% modeling quality. When I scroll through the chart and have M15, it will only go back 6 months.

I am hoping that the bad results are the result of no 15 min data, but in the history center I believe I downloaded it? Why would the last 6 months show 50% modeling quality and the last 2 years show 90%? Is there something I am not doing right to get it to sufficently backtest over that long of a time period, are my 6 month results valid with 50.38% quality results?

 

don't rely on the MT terminal for generation of tickdata for testing. it's crap. unfortunately there's no way around writing your own tickfiles for testing. look up Birt's EA review website (google those words). i warn you, it's not an easy task. but for a professional strategy you can not rely on the terminal generated tickdata, it will not work.

if you are not a coder you need to look for one. there's no way around it, sorry.

regards

Reason: