o.0 Are the 1M High, Low data on MT5 Backtest the same of real Market? I am trying to fix my EA by not using trailing stop loss and use moving stop to Candle 1 High / Low on 1M Chart.
Hi all,
I had a Breakout idea, I create an EA using a 3d Party Software because I don't know anything about coding, the EA working good without any error, back testing on MT5 was great, I did also a MT4 EA and I bought a Tick data software and I tested the EA on MT4 but the result was very very very bad :(
after I googled about MT5 Tester I understand it use the 1M data (Open, Close, High and Low) and I understand that the problem was in the trailing stop, the EA use 3 Pips Trailing stop, on real market and on tick back test it's win only 1 - 3 pips but in normal back test it's win 5- 10 Pips, so I understand that the real market if go up one pip it will fail 3, crazy moves and the trailing stop will not help, the losses will be more bigger than the profit.
Last think I did it, I removed the Trailing stop and made the EA move the stop loss to Candle 1 High if sell / Low if buy ( I will not care what will happen inside the candle not like Trailing stop), but now the EA use multi time frame and I can't / don't know how to do multi time frame back test on MT4, but I did on MT5 and the result was very good.
Now, can I trust MT5 Tester or I must find way to test the EA on MT4 with tick data?
Regards
I hate to burst your buble, but if you have very limited coding skills and are using 3rd party generator, then Testing with or without reliable history or tick data, is the least of your problems.
Also, trailing at 1-3 pips is only profitable on very good ECN brokers with low latency internet connections. Any little slippage, spike or network delay will cause havoc with such a strategy, not to mention the Spread alone will wipe your profit away in an instant!
A good EA needs to be created "by human hand" with knowledge and skill! Every Code Generator I have seen is problamatic with bad and slow code - not at all recomended for such a strategy.
Before you worry about the "quality" of the testing, you should worry about the "quality" of the EA code first!
I hate to burst your buble, but if you have very limited coding skills and are using 3rd party generator, then Testing with or without reliable history or tick data, is the least of your problems.
Also, trailing at 1-3 pips is only profitable on very good ECN brokers with low latency internet connections. Any little slippage, spike or network delay will cause havoc with such a strategy, not to mention the Spread alone will wipe your profit away in an instant!
A good EA needs to be created "by human hand" with knowledge and skill! Every Code Generator I have seen is problematic with bad and slow code - not at all recommended for such a strategy.
Before you worry about the "quality" of the testing, you should worry about the "quality" of the EA code first!
I hate to burst your buble, but if you have very limited coding skills and are using 3rd party generator, then Testing with or without reliable history or tick data, is the least of your problems.
Also, trailing at 1-3 pips is only profitable on very good ECN brokers with low latency internet connections. Any little slippage, spike or network delay will cause havoc with such a strategy, not to mention the Spread alone will wipe your profit away in an instant!
A good EA needs to be created "by human hand" with knowledge and skill! Every Code Generator I have seen is problamatic with bad and slow code - not at all recomended for such a strategy.
Before you worry about the "quality" of the testing, you should worry about the "quality" of the EA code first!
I do agree. Just a little remark, about code quality from an EA generator, it is not a rule of thumb that the code is "bad and slow". For example the MQL5 Wizard certainly provide quality code. But the problem remains, I strongly doubt any generator will ever be able to produce a profitable EA. As you said, it needs "human hand" or rather human brain.
Thank you both for the quick answers, I know it's bad that I can't write my EA with my self, but I did my best for this EA:
Server with Quad core Xeon - 8GB RAM - 160GB SSD - 100MB/S Internet speed - Windows Server 2016:
Brokers: I tried 4 ECN Brokers till I found one with minimum Slippage and here are live trades on live account for last days but with Trailing stop, I had only 1 pip slippage on 1 trade.
and here you can see where is the problem:
about the coding, I already tried some programmers but their copies were very slow on testing but the EA which I did it faster that others.
any suggestions?
Regards
I also want to inform you that this EA is too sample, no indicators just number of candles High and Low, I also tried to make it fast by using (if no trade, one per bar and ones a day) I sow big different in performance after I used this for unnecessary things in the EA.
Can I start think about the back test now?
Regards
Server with Quad core Xion - 8GB RAM - 160GB SSD - 100MB/S Internet speed - Windows Server 2016:
Brokers: I tried 4 ECN Brokers till I found one with minimum Slippage and here are live trades on live account for last days but with Trailing stop, I had only 1 pip slippage on 1 trade.
and here you can see where is the problem:
about the coding, I already tried some programmers but their copies were very slow on testing but the EA which I did it faster that others.
any suggestions?
Yes, you are correct! There are many bad, low quality "human" coders out there which are much worse than the code generators. I agree, but I also said that it requires "knowledge and skill" which means that you have to be very selective with the coders you hire, and the good ones are never cheap.
As for the testing in MT4 and MT5, both can offer "quality" testing but it also requires "knowledge and skill" to back-test, because each environment has its own quirks and limitations that need to be considered. Not being able to code, not only limits your EA coding but it also limits your back-testing quality, because you lack the knowledge and skill to understand how things work "under the hood" and how it affects the test results.
This is why you are having different results in the back-tests, in MT4 with tick data, with MT5, and other scenarios. It boils down to the lack of skill and experience with back-testing and how all the little aspects that can affect the outcome and how to properly set it up correctly.
However, you can do forward tests instead, but it also means that you have to apply statistical analysis of your trades in order to be able to see if it works or not. You have to be able to calculate all those metrics, such as Profit Factor, Win Rate, Risk:Reward ratio, Draw-down, Recovery, and others. So, you will also need knowledge and skill to evaluate your Forward tests as well.
In essence, be it in coding, be it in back-tests, be in forward testing, be it in Live trading - you need to develop your knowledge and skills with more study and research.
Since you are limited by not knowing how to code, try to first develop your Manual trading skills. Search and develop trading strategies that favour manual trading.
If you really want to use EA's, then make sure you use quality human coders, but they will be more expensive. However, you will still need to develop you skills and knowledge at back-testing if you go the EA route (or hire someone that is also highly skilled and experienced, to do back-tests for you)!
- www.metatrader5.com
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hi all,
I had a Breakout idea, I create an EA using a 3d Party Software because I don't know anything about coding, the EA working good without any error, back testing on MT5 was great, I did also a MT4 EA and I bought a Tick data software and I tested the EA on MT4 but the result was very very very bad :(
after I googled about MT5 Tester I understand it use the 1M data (Open, Close, High and Low) and I understand that the problem was in the trailing stop, the EA use 3 Pips Trailing stop, on real market and on tick back test it's win only 1 - 3 pips but in normal back test it's win 5- 10 Pips, so I understand that the real market if go up one pip it will fail 3, crazy moves and the trailing stop will not help, the losses will be more bigger than the profit.
Last think I did it, I removed the Trailing stop and made the EA move the stop loss to Candle 1 High if sell / Low if buy ( I will not care what will happen inside the candle not like Trailing stop), but now the EA use multi time frame and I can't / don't know how to do multi time frame back test on MT4, but I did on MT5 and the result was very good.
Now, can I trust MT5 Tester or I must find way to test the EA on MT4 with tick data?
Regards