I tried backtested on my EA using mt4 , but I get 2 different results .
I bactested using the same EA on market day and get 1 result.
then I backtested again using same EA on no market day and get different results from market open day.
I hope to get opinion from u all whether u all faced this kind of problem before?
And which results is the accurate one? Market open or Market close?
Thanks
Different days have different spreads which is one of the reasons to disregard backtesting results, unless you use proper tick data.
is it during weekend , the results are not accurate? Because I search some forum and they did mentioned about backtesting during weekend too.
Because I backtested on Thursday and Friday. Then I recorded all the data, but then I tried it again on Saturday, the results are totally different. Is it weekend problem or my backtest EA problem?
If you don’t use tick data your MT4 platform will use the current spread (weekend spread). The problem should go away come Monday.
IMHO its not just the weekend your data is just not up to par. What’s happening is you have a variable spread. So whenever you backtest its going to be wrong. Why? Because at each moment in history the spread is different and you don’t have those values (unless you have set bid/ask tick data). your platform is assuming the current spread is the same as all spreads throughout history, which is not true. So in the end everything you’re doing is going to be off, because the results are based on a fault assumption. the 1st most important piece of backtesting is solid data, garbage in garbage out.
MT4, on its own, only keeps open, high, low, and close BID quotes on 1M time frames. It doesn’t include the ASK quote, and thus, has to assume what the spread might be. It does this assumption by using the current spread you see in the Market Watch window.
So if the last quote just before the market closes for the weekend is a high spread (and it usually is,) then your testing over the weekend will assume that spread all the time.
Since MT4 doesn’t use data below 1M out of the box, and has to ‘guess’ things like spreads and the tick activity within that 1M bar (it actually guesses this by looking at the tick volume of the bar and the bar’s range then randomly ticking up and down in a way that fills out the candle), then you’re simply not going to get an accurate backtest on strategies that are sensitive to such short term movement.
To do better, look into tick testing; there’s a few tools out there that modify or work within MT4 to allow tick testing.
Thanks for all the reply. So the conclusion will be, I still will get different results everyday?
I did record down the results of my opening EA for May, then I did some tickdata backtest, the problem is the tick data results don’t fit my recorded results (which is real time trading results).
So I don’t know what should I use to backtest instead of using the data provided in mt4 chart.
My EA is just a normal EA, which using the London Open trading method. Any suggestion?
Why did you assume that backtesting would provide identical results to live trading? In addition to variable spreads you have to contend with slippage and slow connections in live trading. Backtesting has some limited usefulness if you understand how it works. But your expectations don’t seem reasonable to me.
Because I recorded down all the results from my EA on live trading. Then I did the backtest on that particular month and I notice the results from backtest completely alike with my recorded results.
Why don’t you share the EA and we can compare backtest results?