back test and real
-
hello sir I make a very simple ea. it is back test very good result .but I checked real 3 days it is bad result I don't no why.pls help me thanks
;;;;;;;; spread no problomone month profit back test
-
Well, it shows that the modelling quality is 25%, which is bad. Even "Every tick" is in some cases too bad for realistic results - when you have short stops, short trailing stop and things like this.
-
Backtesting is almost all the time crap. Take a look at what I can do with back testing, it's 216 millions profit with 50$ deposit...
Real spread, I can run an EA with 0.1 spread and make billions of $ in one year, but live, it blows out the account.Take time, lose money and work hard with fxdreema. For me, it's the only thing that have worked to finally be profitable. I am still working hard to understand and be able to work with this software.
......
GOLD.jpg -
Well, to decide how accurate the backtest is, it's much better to know how the Tester works. The EA that is generated with fxDreema (and any EA in general) obviously work within that Tick event, but it's the Tester that generates those ticks out of M1 or whatever timeframe is available.
new69mart, you obviously "hacked" the Tester. Your EA actually exploits some weakness of the Tester. But it looks that at least it's clever enough to tell us what is the Modelling quality. 39% is not a big number, only this number alone tells us that the result is not accurate. But I can't fix this from fxDreema, this is because of MetaTrader
-
I am not sure I understand all that... But for now, I am happy with my trading system, but I see many times each day how I could upgrade it and it creates a lot of questions about EAs!
I just created a new one and I got 90% modelling quality, I am trying that one on demo account, maybe it would be good!
......
back test.jpg -
The lowest timeframe is M1. For each M1 Candle you have 4 price levels - Open, High, Low and Close. There is no data for the ticks inside those candles, nothing more that those 4 prices exists.
The first thing that the Tester does is to generate ticks. It reads the information from M1 (or other timeframes) and based on that information it uses some algorithm to create ticks. By the look of each candle it decides how ticks would travel within that candle. Candle's Volume is probably used to decide how many fake ticks to put in that candle. It's the same as if I give you one candle and ask you what was the path of the ticks inside.
If you have candle with height of 10 pips, then fake ticks will be generated for the path of those 10 pips within that candle. But what if you create a trade with stops of 5 pips? If that trade is created and closed within that candle, how accurate will the result be?
It's another story if you for example create a trade in the first tick of the candle and then close it at the first tick of another candle. Those are the Open prices and they should be accurate. Or if you have stops much bigger than the biggest M1 candle.
The problem with Trailing Stop and similar functions is that in most cases they react on small movements. The price moves with 1 pip and Stop-Loss is set somewhere, which means that Stop-Loss is also fake, because it is based on fake ticks.
Here is more to read:
https://www.mql5.com/en/articles/1511
https://www.mql5.com/en/articles/75 (this is for MT5)
