If you have an EA that opens/closes positions on a open candle. How reliable can its back test be with 90% modellig quality?
Thanks
Thanks
FX Dealer Performance 92 replies
Performance statistics 4 replies
Question: evaluating performance 1 reply
Performance, Fortune,& the U.S. President 0 replies
FXWINTRADES managed fund performance 3 replies
DislikedIf you have an EA that opens/closes positions on a open candle. How reliable can its back test be with 90% modellig quality?
ThanksIgnored
DislikedYou may want to keep up with this site. I stumbled upon it a few weeks ago and it'll be interesting to see how it turns out...
http://www.fxtradeblog.comIgnored
DislikedAs long you have the M1 data for your EA, it will be as accurate as you can get if you're using MT4 to test your EA.
Doesn't matter when you open the trade.
GFIgnored
DislikedGirlFlyer thanks for the reply, I know your EA uses M1 Data, and I wanted to ask you if there is a way to get better then 25% modelling quality using the M1?Ignored
DislikedNo, if you do your backtest using M1 setting, 25% is as high as it goes. It's just the way the formula is calculated. I don't know why they made their formula that way, but it is. While you may think 25% is bad, on that setting you have to ignore it. M1 data is as good as it gets (but it still doesn't mean the backtest will be perfectly accurate). I generally find that when I backtest my EA for the same week that I forward test something, it comes pretty close to the same results, so its not a bad deal.
Cheers
GFIgnored
DislikedI disagree. If your strategy uses M1 data and trades within an interval, the back test results will be the least reliable. The bigger the time compression you use, the less dependent it is on the "simulated" tick data being accurate. This is why you get a modeling quality of only 25% when using the minute charts. You will get much more realistic results if you shift 1 period (trade on completed intervals). In this case, you can ignore the 25% modelling quality.Ignored
DislikedI wasn't disagreeing on that point. I have no idea how MT4 calculates the modelling quality and it does seem that 25% is as good as it gets on the M1 data. Keep in mind though, if you are trading on tick data in any time compression it is probably using the same "simulated" tick data that it generates for the M1 data. So I'd expect the same trades at the same time. The question is, which modeling quality is more accurate, 25% or 90%. In my experience, no matter which time compression is used, you just can't rely on the backtest results if you are trading per tick. As GirlFlyer suggested, run it on a demo for abit. A good strategy will work both ways.Ignored