Different behaviour with MQL4 code

I have generated a MQL4 EA from the attached strategy.

It behaves differently on MT4 compared to SQ, although the same CSV tick data (over 4 years) have been used on both sides:

* the number of trades is significantly different : 1600 with SQ, 80 with MT4

* the SL/TP are different: 100 or 110 pips in SQ, 300 pips in MT4

* the volumes are different (logically): 1.0 in SQ, 0.4 in MT4

* the actual profits/loss are slightly different also: 90/100 with SQ, 115 with MT4

* the final MT4 result of course is much less winning than the SQ result

Playing with the input parameters ProfitTargetPips, StopLossPips, MinimumSLPT, MaximumSLPT did not help much.

There must be something I’m doing wrong, or a parameter to fix somewhere, but I could not nfind it.

Any idea ?

Hello, will check the strategy whether a issue can be found. Did you use data from Tickdownloader or any other?

Hello Tomas,

Yes the data comes from Tickdownloader.

Hi

Any news about this issue ?

It’s very  frustrating to have spent quite a lot of time optimizing a strategy and not be able to use it.

Do you get the same behaviour with MT4 ?

Thanks for any help you could provide.

Hi,

I can’t help you, but I can tell you I have a very similar problem. The results in SQ are sometimes extremely different from the backtest in the Metatrader…..but not always……. So far nobody could tell me why! I noticed the same problem with SQ X as well.
The backtestdatas also come from Dukascopy, which makes the whole thing even more confusing……….

Hi,

In my experience, I alwasys Backtest in MT4 and SQ3 (haven’t installed SQX yet) as part of my Workflow; gives me piece of mind’. Well, Never ever both strategies backtests match to each other. However I’m satisfied if ‘more or less’ both Equities look like the same.

I spent quite time trying to figure out what was going on. I saw; Strategies on 4H timeframe had Equities quite fine tuned. Strategies on 1H were more or less similar, and 15M were too much different to each other for my taste

Same issue

Thanks for your feedbacks.

So there is indeed an issue on SQ side. It really looks like a bug in MQL4 translation. If not, it’s a strong limitation that should be documented and made clear before we buy this software.

We can probably not expect an exact reproduction of SQ results in MT4 backtests, although when we use the same input data it should be quite close, but the strategy should at least find a similar amount of signals and put orders with similar TP/SL/volumes. In my case, instead of a reasonable regular winning curve, it becomes completely erratic.

@tomas262 have you found something ?

@Nicolas:

Have you ever checked the problem in SQX?

Hi,

@Marcel. In my case, when I see the long buggy list and time spent on the SQX development I feel have neither time nor energy to deal with more checkings. I prefer to wait for the SQ team to do their job and give us a final product hopefully reliable, stable and bug free.

@Nicolas. I wouldn’t say that the equities are erratic, shouldn’t be. I attach a couple backtest for year 2017. Both cases strategies I have in Real. The equities are not the same; Yeah, it sucks. But are not erratics.

I forgot! Both examples are on 1H. When you go on lower timeframes, in that case it is totally erratic

to Nicolas – your data for GBPJPY is set WRONG.

if i look into the list of trades of your strategy i see, that you have only 2 decimals for GBPJPY – your broker has 2 decimals, or 3?

Because most of the brokers has now for GBPJPY 3 decimals and the price is something like this 148.319

and the second wrong thing, is the pip values – 1. trade – move short from 175.8 to 175.7 is not 100 pips, but only 10 pips

so the backtest is totally wrong and you cant compare it

second thought: the strategy is totally overoptimised – it will fail in real

set for ***JPY pairs correct data – point value 890, piptick size 0.01, piptick step 0.001, spread 6 for example

this is the correct BT

about the comparation between backtests – SQ vs TDS. In our community we are doing monthly comparation between these backtests and comparation from the real accounts.

I must say, that if you dont use pivots and exit rule in building blocks, the backtest are “the same” – there could be differences, but the eq curves need to go the same way, because real acc is another environment (spreads, slippages, gaps, etc.)

we are using strats for M5-H4 and its not true, that the comparation of BT for lower TF is worse

this is some comparation curves between backtests and between reall acc and BTs

@Marcel: I have not tried SQ yet, I may take a look

@Enric: in my case it’s 1H too, I’ll see how it goes with a fixed setting

@hankeys: thank you for the analysis, my broker has indeed 3 digits for GBPJPY, I have to check and fix the settings